CLAUDE A. NOZÈRES

RÉGIME ALIMENTAIRE DU BÉLUGA, DELPHINAPTERUS LEUCAS, de l’estuaire du Saint-Laurent, Canada,

tel que révélé par l’analyse des acides gras du lard

Mémoire présenté à la Faculté des études supérieures de l'Université Laval

dans le cadre du programme de maîtrise en biologie pour l’obtention du grade de maître ès sciences (M. Sc.)

DÉPARTEMENT DE BIOLOGIE FACULTÉ DES SCIENCES ET DE GÉNIE

UNIVERSITÉ LAVAL QUÉBEC

JUILLET 2006

© Claude A. Nozères, 2006

2

Résumé

Une connaissance du régime alimentaire des espèces constitue un élément clé pour

la compréhension de l’écologie des espèces. Or, peu de techniques existent afin d’examiner

le régime alimentaire contemporain du béluga du Saint-Laurent étant donné le statut

précaire de cette population et la rareté de contenus stomacaux chez les individus retrouvés

morts. L’analyse des profils d’acides gras dans le lard et dans leurs proies potentielles

constitue un outil susceptible de fournir certains indices concernant leur régime alimentaire.

C’est donc dans ce but qu’une étude comparative des profils d’acides gras a été menée.

Dans une première étape, le profil d’acides gras a été examiné chez environ 60 espèces de

proies potentielles du béluga provenant de l’estuaire et du golfe du Saint-Laurent. Cette

analyse a révélé la nécessité de considérer les effets des lipides totaux, de la taille, des

saisons et des lieux d’échantillonnage lors de l’examen des profils d’acides gras des espèces

afin d’améliorer le couplage des profils des proies potentielles avec ceux d’un prédateur tel

que le béluga. Les profils d’acides gras du lard ont ensuite été comparés entre les bélugas

et quatre espèces de pinnipèdes qui se trouvent dans l’estuaire et le golfe du Saint-Laurent.

Les espèces de même que les sexes et classes d’âge ont généralement pu être distingués sur

la base des profils d’acides gras. Une gamme d’analyses multivariées ont révélé des

différences qui semblaient liées à des acides gras d’origine alimentaire, confirmant ainsi le

potentiel de ces techniques dans l’analyse comparative de la diète des mammifères marins.

La caractérisation des signatures d’acide gras des proies potentielles et des prédateurs

constitue une étape essentielle dans le cheminement menant à une meilleure compréhension

de l’écologie alimentaire de ces animaux.

ii

Abstract

Knowledge of diet is an important element for understanding the ecology of species.

Currently, there are few techniques to examine the contemporary diet of St. Lawrence

belugas considering the precarious status of the population and the rarity of prey remains in

the stomach of stranded individuals. Fatty acid analyses of prey and blubber may provide

the tools to indicate diet. It was with this purpose that a comparative study of fatty acid

profiles was undertaken. As a first step, about 60 potential prey species for belugas were

collected in the Estuary and Gulf of St. Lawrence and were examined for their fatty acid

profiles. The analysis revealed the need to consider the effects of lipid content, size,

season, and sampling location when evaluating species fatty acid profiles in order to

improve evaluations of prey profiles of a predator such as the beluga. Blubber fatty acid

analyses were subsequently compared between belugas and four species of pinnipeds that

are encountered in the Estuary and the Gulf. Species along with other classes such as sex

and age could be distinguished on the basis of their fatty acid signatures. A suite of

multivariate analyses confirmed the role of diet-linked fatty acids in differentiating the

groups, thereby confirming the potential for these techniques to conduct a comparative

analysis of diets among these marine mammals. The characterization of fatty acid

signatures of potential prey and predators represents essential steps on the path towards a

better understanding of trophic ecology in these animals.

Avant-propos

À chacune des étapes, ce projet ne se serait pas réalisé sans les contributions de beaucoup

de monde, mais je dois donner des remerciements particulièrement profonds à mes

superviseures.

Avant tout, à ma co-directrice, Dr Véronique Lesage, de l’Institut Maurice-Lamontagne

(IML), de Pêches et Océans Canada, Région du Québec, qui est la raison d’être, autant que

le ciment qui permettait à tous ces liens et ces opérations d’exister. C’est grâce à sa

grandeur d’esprit que mon projet a fonctionné et dépassé tout obstacle, grand et petit. Je

suis vraiment chanceux comme étudiant de la connaître comme amie en même temps que

‘boss’! Si ma réussite se fait remarquer, il faudra considérer ses talents de chercheuse

autant que son amitié afin de comprendre la source.

À ma directrice, Dr Helga Guderley de l’Université Laval, pour son soutien et sa patience.

Elle est une chercheuse exemplaire dont la sagesse en même temps que la bonne humeur

sont des sources de plaisir pour tous ceux qui ont eu la chance de la rencontrer.

To Dr. Sara Iverson (Dalhousie University), for accepting to collaborate in the project, thus

making it all possible in the first place! Thanks also for providing timely advice and

inspiration, and for setting an example of scientific scholarship, through your seminal

publications and presentations in the intriguing field of lipid physiology and ecology.

iv Thank you to Shelley Lang, and all the technicians of the ‘Fat lab’ at Dalhousie, for the

thousand-plus chemical extractions analyses they performed over the course of my project.

Merci aux échantillonneurs du MPO et MAPAQ pour leurs conseils et contributions,

incluant tous les membres des équipages du NGCC Calanus II et du NGCC Alfred Needler

(2000 - 2002).

Merci aux Drs Ladd Johnson et Jacques Larochelle pour leur soutien au comité et les

conversations fulgurantes dans les cours d’écologie et physiologie. Merci aussi au Dr

Larochelle d’inciter les étudiants à réfléchir entre les domaines de l’écologie et de la

physiologie, ce qui a en partie inspiré ma conclusion ici.

Merci à tous mes amis de l’IML, surtout à Diane Archambault, Denis Chabot, Louise

Savard, Lyne Morissette et Yves Morin pour leurs conseils bien appréciés. Et surtout à

trois amis spéciaux, qui, comme Véro, m’ont supporté (et même toléré) avant et durant ce

projet: Pierre Carter, Jean-François Gosselin et Mike Hammill.

Dès mon arrivée à Québec, tout le monde me tendait une main de bienvenue qui ne lâchait

pas, ce qui ne cessait pas de me surprendre. Si jamais je suis regardé comme un type

correct, c’est en espérant suivre un peu leurs exemples de professionnalisme autant que

d’amitié. Mais mon chemin a commencé sur une plage à Vancouver, et pour tous ces

souvenirs et les voyages, je tiens fort à remercier mes parents, Claude et Marguerite.

Aux enfants

vi

Table des matières Liste des tableaux................................................................................................................ viii

Liste des figures .................................................................................................................... xi

Chapitre 1 : Introduction générale ......................................................................................1

1.1. Contexte ....................................................................................................................2

L’écologie trophique régionale .......................................................................................2

L’écologie alimentaire des mammières marins ..............................................................5

1.2. Objectifs généraux ..................................................................................................12

Chapitre 2 : Among- and within-species variability in fatty acid signatures of fish and

invertebrate prey of the St. Lawrence beluga whale.............................................................20

Résumé..............................................................................................................................21

Abstract .............................................................................................................................21

2.1 Introduction..............................................................................................................22

2.2. Materials and methods ............................................................................................27

2.3. Results.....................................................................................................................32

Representation in prey base ..........................................................................................32

Lipid content .................................................................................................................32

General fatty acid compositions....................................................................................33

Multivariate comparisons among and within species ...................................................34

Digestive tract contents.................................................................................................41

2.4. Discussion ..............................................................................................................42

Species sampling...........................................................................................................43

Lipid contents................................................................................................................46

vii

General fatty acid comparisons.....................................................................................48

Inter- and intra-specific fatty acid variability ...............................................................49

2.4. Conclusion ..............................................................................................................57

Chapitre 3 : Comparing fatty acid signatures among marine mammals of the Estuary and

Gulf of St. Lawrence: an evaluation using multivariate methods........................................91

Résumé..............................................................................................................................92

Abstract .............................................................................................................................92

3.1. Introduction.............................................................................................................94

3.2. Materials and Methods............................................................................................97

3.3. Results...................................................................................................................101

General patterns in the abundance of the different fatty acids....................................101

Classification models ..................................................................................................102

3.4. Discussion .............................................................................................................109

3.5. Conclusion ............................................................................................................126

Chapitre 4 : Conclusions générales...................................................................................147

4.1. Sommaire des résultats..........................................................................................148

4.2. Approches futures .................................................................................................150

Bibliographie Générale.....................................................................................................156

Annexe I : Examination of extra fish specimens for digestive tract contents ....................191

Annexe II : List of available potential prey species and spatiotemporal analyses............194

Annexe III : Extra figures from Chapter 2 ........................................................................196

Annexe IV : Alternate presentations of figures in Chapter 3 ............................................203

viii

Liste des tableaux

Table 1.1. Liste des acides-gras fréquemment détectés, organisés par le nom chimique, la

longueur de chaîne, et le degré d’insaturation. .............................................................15

Table 1.2. Sommaire des techniques statistiques multivariées fréquemment employées

lors d'analyses des données biochimiques. ...................................................................16

Table 2.1. Summary of samples by species, season and sub-region: Upper Estuary (A),

Lower Estuary (B), Sept-Iles (C), and northern Gulf (D). Seasons are spring (April–

June), summer (July–Aug.), autumn (Sept.–Oct.) and winter (Feb.). Samples were

analysed either as whole specimens or as a pooled set of specimens in the case of

zooplankton (amphipods, copepods, euphausiids)........................................................59

Table 2.2. Composition (% of total by weight) for 25 fatty acids, each averaging

>0.4%, among all species groups analysed ( = fatty acid not retained in statistical

analyses). Shad, herring, and salmon were each segregated into small (<18 cm) and

large (>18 cm) groups. Specimen standard length, wet weight, and lipid content are

also presented. Values are means ± SEM …………………………………………...62

Table 2.3. Principal component (PC) analysis for all specimens (n = 1029) using the set

of 16 logratio fatty acids. Summary of significant (> |0.5|) factor loadings for the

retained PCs (eigenvalues >1) following Varimax rotation. Maximum absolute

loadings are indicated in bold. ......................................................................................76

Table 2.4. Principal component (PC) analysis for eight repeated species in autumn using

the set of 16 logratio fatty acids. Summary of significant (> |0.5|) factor loadings for

the retained PCs (eigenvalues >1) following Varimax rotation. Maximum absolute

loadings are indicated in bold. ......................................................................................77

Table 3.1. Marine mammals sampled in the Estuary and Gulf of St. Lawrence between

1996-1998……………………………………………………………………...........128

ix Table 3.2. Fatty acid compositions of five marine mammal species sampled in the

Estuary and Gulf of St. Lawrence. Values are means by mass % (± SEM) for 33 fatty

acids with overall mean abundances ≥ 0.2%. The 18 fatty acids selected for use in

statistical analyses are indicated in bold. ....................................................................129

Table 3.3. Significant loadings (>0.5) of 17 fatty acid logratio variables on the four

principal components (PC) that were retained (Eigenvalues > 1). The maximum

loading for each fatty acid variable is highlighted in bold. Also shown is the % of

variance (Varimax-rotated) accounted for by each PC. Kaiser-Meyer-Olkin measure

of sampling adequacy was 0.772. ...............................................................................130

Table 3.4. Results of one-way ANOVAs (F4,138, P < 0.005) for differences among species

groups for each PC factor. Values in the same column not sharing a common

superscript letter are significantly different (post hoc Tukey HSD multiple comparison

of means, overall a = 0.05). Species groups are: hooded seals (Cc), beluga whales

(Dl), harp seals (Pg), grey seals (Hg), and harbour seals (Pv). ...................................131

Table 3.5. Results of ANOVA (F8, 134, P < 0.001) to test differences among clusters on

each PC factor. Mean values in the same column not sharing a common superscript

letter are significantly different (post hoc Tukey HSD multiple comparison of means,

overall a = 0.05). The number of members for each cluster is indicated following each

species: hooded seals (Cc), beluga whales (Dl), harp seals (Pg), grey seals (Hg), and

harbour seals (Pv)........................................................................................................132

Table 3.6. Means (±SEM) for the subset of 18 fatty acids (normalized over 100% for each

individual) on the nine marine mammal cluster groups resulting from the hierarchical

cluster analysis on 4 PC factors derived from the 17 logratios...................................133

Liste des figures

Figure 1.1. Schéma d’une acide gras typique, avec un carbone estérifié en tête (COO-) et un

groupe méthyle (CH3) terminal. L’exemple présenté ici concerne une chaîne saturée

(i.e., pas de lien double entre les carbones) avec le 12e carbone en position terminale

(ω)…………………………………………………………………….……………….17

Figure 1.2. Exemple de spécificité des voies de synthèse pour les acides gras à longue

chaîne polyinsaturée. Les flèches solides indiquent les voies retrouvées chez les

mammifères tandis que les flèches pointillées représentent les voies exclusives aux

autres types d’organismes. Tiré de la Fig. 1 de Leonard et al. (2004)..........................18

Figure 1.3. Schémas typiques de l’analyse des données d’acides gras à l’aide des

techniques multivarieés semblables à celles utilisées au chapitre 3 : a) démarche lors

d’une analyse suivant un arbre de classification. b) démarche lors d’une analyse

impliquant une suite d’analyses en composants principales, de groupements, et des

fonctions discriminants. Les encadrés gris illustrent des exemples de sorties

graphiques associées avec ces techniques de classification, soit; c) séparation en arbre

de décisions, d) projections linéaires, et e) association de cas lors d’analyses de

groupements.. ................................................................................................................19

Figure 2.1. a) Study area (box) and b) sampling zones in the Estuary and Gulf of St.

Lawrence, Canada.. .......................................................................................................78

Figure 2.2. Selected species with significant amounts of the nominally ‘trace-level’ fatty

acid 22:1n-7 not used in further evaluations. Dashed line = global mean of 0.27%.

Diamonds are the horizontal spread across the mean (middle lines), with significant

differences between species indicated by non-overlapping vertical apices. .................78

Figure 2.3. Cumulative mean barplots by species for four minor polyunsaturated fatty

acids that were retained for evaluation. Species groups are ranked vertically in

descending order of abundance for the fatty acid 18:2n-6............................................80

xii Figure 2.4. Matrix scatterplot of factor scores on the retained PC Factors 1 through 4,

accounting for 81.7% of total cumulative variance in the 16 logratio fatty acid

variables on all samples. Groups labelled are those that scored farthest from central

majority. ........................................................................................................................81

Figure 2.5. Scatterplot of scores on Factor 1 and 2 retained from a separate PC analysis

conducted on eight species of prey commonly found in the Estuary in autumn.

Individual scores are indicated by species name (a) and lipid content .........................82

Figure 2.6. Original PC analysis showing only the scores for four cod species...............83

Figure 2.7 . Original PC analysis showing only the scores for smelt, capelin, and herring.

Overlapping scores on the first two PC factors (enlarged, in upper right, with convex

area shading according to species) are in contrast to separations seen in the full matrix

with Factors 3 and 4 (lower left). ..................................................................................84

Figure 2.8. Example of within-species variability, with plotted scores on Factors 1 and 2

for all capelin specimens by year, season, and locations: SI = Sept-Iles (N. Gulf), NS =

North Shore (Lower Estuary), SS = South Shore (Lower Estuary), UE = Upper

Estuary.. ........................................................................................................................85

Figure 2.9. Example of between-species seasonal variability, with plotted scores on

Factors 1 and 2 for autumn specimens of smelt (triangles), capelin (diamonds), herring

and sardines (crosses)....................................................................................................86

Figure 2.10. Scores on Factors 1 and 2 for autumn samples weighted by total lipid

content for smelt (orange), capelin (green) and herring (blue) and herring sardines

(cyan)…...……………………………………………………………………………..87

Figure 2.11. Biplot on the first two discriminant functions from an analysis classifying

those prey species sampled in greatest number (range = 29–126), comprising three

pelagic fishes (smelt, capelin, herring), three demersal fishes (flounder, plaice,

tomcod), and one invertebrate (crangon shrimp). Circles represent 95% confidence

xiii

intervals of mean scores by species. Also projected are the more-important

contributions from 11 of the 16 logratio fatty acids used in the classification, along

with individual scores. ..................................................................................................88

Figure 2.12. Biplot on the first two discriminant functions from an analysis classifying

15 species groups (>17 samples each, total n = 699) of potential prey of beluga, using

16 fatty acid logratios as input variables. Circles denote 95% confidence intervals for

mean scores by group (sculpins and shrimps comprised three species each). Non-

overlapping circles indicate significantly different groups. For reference, scores are

plotted for all samples, including those not used in the analysis, such as eels and

whelks (shaded areas). Selected individuals of Greenland halibut are marked as being

distinct from their respective groups. Variables are projected according to their

relative contributions on the two discriminant functions. For clarity, the minor

contributions of fatty acids 18:3n-3, 20:1n-11, and 22:1n-9 are not included..............89

Figure 3.1. Approximate sampling locations in the Estuary and Gulf of St. Lawrence.

Beluga carcasses were recovered along the shores of the lower Estuary (dashed

outline). Seals were from (1) Bic, for harbour and grey seals (Estuary), (2) Godbout,

for harp seals (Estuary), (3) near the Magadalen Islands for breeding harp seals (Gulf),

(4) adjacent to Prince Edward Island (P.E.I.) for breeding hooded seals (Gulf), and (5)

near Port Hood, Nova Scotia for breeding grey seals (Gulf). .....................................134

Figure 3.2. Classification trees for five different species of marine mammals (n = 143)

collected in the Estuary and Gulf of St. Lawrence grouped by species: (a) initial

classification using all available (56) fatty acids, (b) classification produced while

using the reduced set of 18 fatty acids. Fatty acids indicated in the rounded boxes

represent variables and the level (% amount) chosen to split cases. The number of

individuals of each species are listed in the terminal nodes (squared boxes), with the

numerically dominant species in bold and the ‘misclassified’ individuals in normal

type. Species classes are beluga (Dl), harbour seals (Pv), harp seals (Pg), grey seals

(Hg), and hooded seals (Cc). Total misclassification rate was 10 and 11 of 143

individuals for (a) and (b) respectively. ......................................................................135

xiv Figure 3.3. Classification trees by gender for adult male (M) and female (F) beluga (a),

hooded seals (b), harp seals (c), grey seals (d) and harbour seals (e and f) as

determined by CART while using the set of 18 fatty acids. .......................................137

Figure 3.4. Classification trees by age-class (P = neonates; J = sub-adults; A = adults) for

(a) harp seals, (b) grey seals, and (c) harbour seals, as determined by CART while

using the set of 18 fatty acids......................................................................................138

Figure 3.5. Varimax-rotated PC factor scores of the five marine mammal species on a

matrix of of 4 PCs obtained using 17 fatty acid logratio variables, with species groups

as marked. A separate label was added for a newborn beluga (black cross) and three

beluga males (blue halos) that scored away from other belugas and towards the

pinnipeds on PC 1.. .....................................................................................................139

Figure 3.6. Discriminant functions analysis of the five species of marine mammals using

(a) scores on the four factors retained from the principal components (PC) analysis,

and (b) the 17 fatty acid logratios as input variables. Group centroids (star) and 95%

confidence intervals are indicated for each species. Variables (PCs or fatty acids) are

projected (rescaled to 200% for visibility) according to their relative contributions on

the two discriminant functions. The cross-validated classification success rate was

84% and 99% (120/143 and 142/143 individuals correctly classified) for (a) and (b),

respectively. ................................................................................................................140

Figure 3.7. Dendrogram of hierarchical clustering with Ward’s method using the four

principal component factors as input variables. Horizontal scale denotes the increase

in within-cluster’s variance, rescaled from 0 to 25. Vertical dashed line highlights the

joining of cases into a nine-cluster solution. Individual members are summarised in

each cluster by their species: hooded seals (Cc), beluga (Dl), grey seals (Hg), harp

seals (Pg), and harbour seals (Pv). Hooded seals comprised only adult individuals.

Belugas were all adults except for one individual each in clusters 2 and 4. Beluga and

hooded seals are sub-labelled by gender: female (F), and male (M). Other pinnipeds

are sub-labelled by age-class: adult (A), sub-adults (J), and neonates or pups (P). The

xv

highest aggregation into two clusters is suggested by the labelled boxes for a grouping

of belugas and principally sub-adults seals from the Estuary relative to all other

pinnipeds from the Estuary and the Gulf. ...................................................................142

Figure 3.8. Mean diamond scatterplots of scores for all individuals by cluster number on

PC factors 1 through 4 (a-d). Diamonds indicate cluster group means ± 95%

confidence intervals (vertically) in proportion to group size (horizontally). Scores are

marked by species: beluga (open circles), hooded seals (triangles), harbour seals (open

boxes), grey seals (x), and harp seals (crosses)...........................................................144

Figure 3.9. Discriminant function analysis of the classification results for the nine

clusters. Large circles denote the 95% confidence interval centroids of mean scores (=

stars) for clusters 1 through 9. Variables are projected (rescaled to 200% for visibility)

according to their relative contributions on the two discriminant functions. (a) Biplot

of mean cluster scores using the four principal components (PC) as input variables.

The overall classification success rate was 97% (139/143 individuals correctly

classified). (b) Biplot of mean cluster scores using the set of 17 fatty acid logratios as

input variables. The overall classification success rate decreased to 93% (133/143

individuals correctly classified). .................................................................................145

1

Chapitre 1

Introduction générale

2

1.1. Contexte

Un objectif continuel de l’écologie est de progresser de l’observation et de la

description à des explications causales et prédictives (May 1999, Underwood et al. 2000,

Power 2001, Bowen 2005). En ce sens, la maturation des sciences biologiques cherche à

suivre le chemin parcouru par les sciences physiques. Par exemple, les disciples de

l’astronomie ont dû développer les concepts de Copernic (c’est-à-dire le soleil, et pas la

terre, est le centre de notre système planétaire) dans un premier temps et ceux de Kepler

(les planètes ont des orbites elliptiques, pas circulaires) dans un deuxième temps pour

arriver à un ensemble de théories qui constituait un véritable progrès. Antérieurement, les

observations d’orbites excentriques de Ptolémée étaient plus correctes en prévisions que les

héliocentriques de Copernic avant l’appui des lois de Kepler. De la même manière, il se

peut qu’une période de transition soit nécessaire afin de relier les observations accumulées

avec les idées contemporaines en écologie et en biologie des organismes. C’est dans ce

contexte qu’est lancé le défi d’étudier les interactions trophiques entre les prédateurs et

leurs proies à l’aide des biomarqueurs. Ce défi consiste à examiner d’abord la variabilité

spatio-temporelle écologique, puis les raisons pour lesquelles les lipides constitueraient de

bons indicateurs écologiques et enfin la manière dont les méthodes statistiques multivariées

peuvent nous aider à présenter ces informations, de façon à mettre de l’avant des

hypothèses rigoureuses.

L’écologie trophique régionale

L’analyse des écosystèmes est une démarche complexe, ambitieuse et coûteuse à

cause du grand nombre de variables d’intérêt potentiel liant les organismes et leur

3 environnement. C’est pourquoi elle nécessite souvent des études prolongées et des échelles

d’étude variées afin de documenter les patrons et les processus qui les affectent (Werner et

al. 2004, Witman et al. 2004, Anderson et al. 2005). Ces analyses visent non seulement à

dégager des principes généraux liés au fonctionnement des écosystèmes, mais aussi à

discerner les tendances majeures et mineures en faisant des inférences solides, non-biaisées

par des cas spéciaux ou du bruit dans les données (Underwood et al. 2000, Smayda et al.

2004, Bowen 2005, McIntyre et Mckitrick 2005). Une façon d’examiner ces patrons est de

se référer à la variabilité spatio-temporelle pour documenter les différences et les

ressemblances entre les sites et les périodes. Dans ce contexte, il est essentiel de

comprendre le choix des acteurs autant que les variables qui vont servir à retracer les liens

écologiques.

La documentation des interactions trophiques des populations, comme la

consommation ou la compétition pour un niveau trophique ou des espèces-clés, qu’il

s’agisse d’espèces fouragères ou de prédateurs majeurs, est souvent la cible principale des

analyses écosystémiques (Bruno et al. 2003, Fanshawe et al. 2003, Roberge et Angelstam

2004). La révélation des associations avec la géographie, telle que l’hydrologie locale et le

climat régional, est aussi d’intérêt puisqu’elle peut parfois identifier les facteurs régissant

ces liens (Thacker et Lewandowicz 1997, Zamon 2001, Abookire et Piatt 2005). Par

exemple, on peut tenter de déceler si les prédateurs varient leur régime alimentaire en

fonction de l’abondance des proies (l’opportunisme) ou s’ils affichent des préférences

alimentaires (sélection dirigée), comme démontrent les études directes des oiseaux

alimentant leurs petits ainsi que les études indirectes des statistiques de pêche (Wilson et al.

2002, Santos et Pierce 2003, Litzow 2004). Les tendances ainsi dévoilées pourront être

4 utiles pour formuler des hypothèses lors de la reconstruction des interactions trophiques

historiques ou encore pour prédire les interactions trophiques futures (Pitcher 2001, Pauly

et Maclean 2003, Choi et al. 2004, Frank et al. 2005).

L’un des risques associés à l’étude des processus et des espèces-clés est la

possibilité de sous-estimer l’importance de certaines composantes moins remarquées et de

les exclure a priori de l’analyse, réduisant ou biaisant de ce fait notre compréhension du

fonctionnement des écosystèmes (Underwood 2000). Deux exemples viennent à l’esprit.

Dans le premier, on étudie l’écosystème dans le contexte des transferts de biomasse tel que

celui du carbone entre certains groupes taxonomiques, ignorant le rôle des composantes ou

des éléments mineurs, ou encore celui des classes de taille ou d’âge qui sont égalements

importantes dans le contexte des interactions trophiques (Jennings et al. 2002, Birkeland et

Dayton 2005, Woodward et al. 2005). Malgré les critiques potentielles associées a une telle

simplification, cette approche visant à modéliser selon un processus clé (le transfert de

carbone) a connu beaucoup de succès dans le cadre d’études régionales, réalisée

généralement à l’aide du logiciel ‘ECOPATH’ (Christensen et Pauly 1992). Un deuxième

cas de biais possible des études écosystémiques est un intérêt démesuré pour certains

vertébrés de grande taille tels que les oiseaux, les poissons commerciaux et les mammifères

marins, non pour leur importance écologique mais pour leur caractère charismatique ou leur

intérêt pour les humains (Schmelzer 2001, Sergio et al. 2005). Plusieurs espèces

d’invertébrés et de petits poissons sont méconnues, n’ont pas de valeur commerciale et sont

souvent peu remarquées ou échantillonnées par les engins de pêche, malgré leur présence et

leur importante contribution au régime alimentaire de prédateurs tels que les oiseaux et les

mammifères marins (Bryant et al. 1999, Hjelset et al. 1999, Payne et al. 1999, Santos et

5 Haimovici 2001). Les amphipodes (Themisto spp.) et le lançon (Ammodytes spp.) sont des

exemples de ce phénomène, en tant que source importante de nourriture pour les prédateurs,

mais leur biologie et leur abondance demeurent encore peu connues (Nilssen et al. 1995,

Bowen et al. 2002, Dunham et Duffus 2002).

L’écologie alimentaire des mammières marins

L’estimation de la disponibilité des proies est d’intérêt pour plusieurs questions

fondamentales liées à la gestion des populations de mammifères marins (Shelton et al.

1997, Flaaten 1998, Bundy 2001, DeMaster et al. 2001, MacKenzie et al. 2002, Read and

Brownstein 2003, Pitcher 2004).

Les populations les plus abondantes, surtout celles de pinnipèdes, attirent l’attention

de par leur potentiel de nuire à la pêche et ce, particulièrement dans les régions où les

stocks de poissons commerciaux sont en déclin ou en mauvaise condition (Trites et al.

1997, Swain et Sinclair 2000, McLaren et al. 2001, Yodzis 2001, Pauly et Maclean 2003).

Par contre, plusieurs espèces de cétacés ont été surexploitées historiquement, au point où la

taille de leur population a subi des baisses importantes (Mitchell 1974, Reeves et al. 1999,

Nichol et al. 2002, Roman et Palumbi 2003, Baker et Clapham 2004). Malgré les mesures

de conservation des récentes décennies, plusieurs populations de cétacés semblent croître à

des taux sub-optimaux (Caswell et al. 1999, Anderson 2001, Stevick et al. 2003, Baker et

Clapham 2004). La perception d’un manque de reprise des populations soulève la question

des facteurs pouvant limiter leur rétablissement, incluant la disponibilité et la qualité des

ressources alimentaires.

6

Bien que la nourriture semble l’une des contraintes les plus logiques à la croissance

des populations, sa contribution au statut des espèces demeure l’une des contraintes les plus

difficiles à établir de façon non équivoque (Fritz et Hinckley 2005). Pourtant,

l’identification de changements dans l’assemblage d’espèces de la chaîne alimentaire

pourrait constituer un bon indicateur des problèmes courants ou éventuels (Wanless et al.

2005). À cet égard, l’état des populations de mammifères marins pourrait servir

d’indicateur de l’état des écosystèmes. Des exemples de monitorage de communautés

utilisant comme indice les populations d’espèces en péril incluent les études des otaries de

Steller en Alaska (Merrick et al. 1997, Trites et Donnelly 2003, Fritz et Hinckley 2005), des

baleines franches de l’Atlantique Nord (Greene et Pershing 2004) et des baleines grises du

Pacifique Nord (Perryman et al. 2002).

Qu’ils soient en croissance ou menacés d’extinction, il s’avère prudent d’intégrer

ces prédateurs du sommet de la pyramide trophique dans tout discours visant les

interactions trophiques, compte tenu des biomasses importantes de proies qu’ils ont le

potentiel de consommer (Kenney et al. 1997, Schweder et al. 1998, Pauly et Maclean 2003,

Trites et al. 2004). Toutefois, la difficulté de mener des études dans le milieu aquatique fait

en sorte que notre connaissance de l’écologie alimentaire de ces animaux reste souvent

incomplète. Une exception importante à cette règle de méconnaissance est le cas des

loutres de mer (Enhydra lutris). Le déplacement relativement restreint de ces mammifères

marins et de leurs proies benthiques qu’ils consomment lorsqu’ils sont allongés à la surface

facilite leur étude. Ainsi, l’impact de ces prédateurs sur certaines communautés marines

spécifiques a pu être évalué (Estes et Duggins 1995). Par contre, la plupart des

mammifères marins ont une tendance à parcourir de longues distances et à s’alimenter sous

7 l’eau où ils sont difficiles à repérer et à observer (Gaskin 1982, Bowen et Harrison 1996,

Bowen 1997, Pauly et al. 1998b, Trites 2003). Le développement de techniques

alternatives pour complémenter l’information obtenue par les méthodes plus traditionnelles,

telles que les observations en surface ou l’examen des traces non-digérées dans les fèces

s’est avéré nécessaire pour approfondir nos connaissances du rôle des mammifères marins

au sein des écosystèmes (Tollit et al. 1998).

Les biomarqueurs du régime alimentaire

Les techniques traditionnelles utilisées en écologie alimentaire comportent certains

biais et peuvent être difficilement applicables à l’étude de certaines espèces (e.g., Pierce et

Boyle 1991). Les techniques d’analyse du régime alimentaire vont de l’analyse des

derniers repas, par l’identification des restes des proies à travers l’examen du tractus

digestif ou des fèces et la reconstruction de la diète, à des méthodes indirectes intégrant le

régime alimentaire sur des périodes plus longues et qui sont fondées sur la proportion

détectées de certains molécules ou composés. Un exemple de ce genre de traceur indirect

du régime alimentaire, ou biomarqueur, et qui est maintenant très répandu est l‘analyse des

rapports de certains isotopes stables tels que le carbone et l’azote dans les tissus des proies

et des prédateurs (DeNiro et Epstein 1981, Tieszen et al. 1983, Lesage et al. 2001). Une

autre classe de biomarqueurs est celle des acides gras, une composante des lipides

remarquée depuis des décennies comme une source potentielle d’identification et de

quantification de certains processus incluant le régime alimentaire des organismes (Ackman

et al. 1971, Ackman 1989, Dalsgaard et al. 2003). Comme pour les études impliquant les

isotopes stables, l’intérêt pour ce type de biomarqueurs s’est accrû avec la disponibilité

8 d’équipement analytique et leur potentiel pour l’emploi de méthodes moins invasives

(Connor et al. 2000, Dalsgaard et al. 2003).

Les écosystèmes marins sont remarquables par leur diversité en lipides biologiques

et en acides gras (Ackman 1989, Zhukova et Aizdaicher 1995, Volkman et al. 1998). Les

acides gras sont composés de chaînes de carbones, dont la longueur et le degré de saturation

(nombre des liens simples et doubles entre les carbones) leur donnent des propriétés

distinctes (Fig. 1.1). Bien que plusieurs acides gras puissent être synthétisés et modifiés par

une variété d’organismes, les acides gras polyinsaturés proviennent typiquement des

végétaux incluant le phytoplancton. De nombreux acides gras (souvent plus d’une

soixantaine) qui composent les lipides peuvent être identifiés lors d’une analyse typique de

chromatographie en phase gazeuse (Table 1.1). Plusieurs de ces acides gras proviennent

exclusivement de la diète alors que d’autres peuvent être produits par l’organisme, mais à

un taux insuffisant pour répondre aux besoins ou pour expliquer les niveaux observés

(Sprecher 2000, Leonard et al. 2004; Fig. 1.2). Ainsi, la composition en acides gras peut

discerner certaines tendances générales, comme l’alimentation herbivore vs carnivore (Auel

et al. 2002), pélagique vs benthique (Graeve et al. 1997), poisson vs invertébré (Phillips et

al. 2003), ou eau douce vs eau de mer (Smith et al. 1996). Il existe également un potentiel

d’élucidation des patrons plus spécifiques allant du groupe d’espèces, surtout pour les

prédateurs sténophages, à l’identité taxonomique des proies ingérées.

Il est possible d’atteindre une plus grande spécificité dans l’identification des proies

lorsque l’ensemble du profil d’acides gras, communément appelé la signature d’acides gras,

est utilisé plutôt que seulement un ou quelques acides gras (Smith et al. 1997). C’est aussi

le cas avec d’autres biomarqueurs tels que les isotopes stables individuels.

9 L’enrichissement ou la déplétion des tissus des prédateurs en regard de certains isotopes

représente une des méthodes répandues pour adresser les questions relatives aux

interactions trophiques (Hobson 1999, Lesage et al. 2001, Jennings et al. 2002, Das et al.

2003). Bien que les isotopes stables de carbone et d’azote soient les plus populaires,

d’autres isotopes tels que le baryum, le mercure, le plomb, le césium, et l’arsenic peuvent

également servir à l’identification de sources environnementales pouvant révéler l’origine

des ressources alimentaires. La combinaison des traceurs atomiques (isotopes) avec les

structures moléculaires (chaînes d’acides gras) constitue un exemple de technique

permettant d’augmenter la spécificité du traçage des sources alimentaires (Stott et al. 1997,

Mottram et al. 2003, Chamberlain et al. 2004). Outre la combinaison de traceurs déjà

communément employés, l’utilisation de nouvelles versions d’analyses chromatographiques

et spectroscopiques promettent d’offrir un rehaussement de la précision analytique (Christie

2003, Takaichi et al. 2003, Wetzel et Reynolds 2004). Cependant, comme toute nouvelle

initiative technologique, le potentiel de ces outils doit se réaliser au coût de l’inexpérience

et d’essais répétés lors de leur développement pour ultimement être utiles dans de

nombreuses études futures. Néanmoins, poussé par les intérêts commerciaux en santé, les

innovations en technologie lipidique ont connu un développement pratique et rapide

(Ackman 2002).

Bien que la continuation du développement des techniques analytiques ajoutera à la

précision des analyses, parallèlement, l’un des besoins actuellement les plus criants est de

gérer les données issues de ces activités. Une meilleure compréhension du rôle de la

physiologie sur le métabolisme des acides gras servira à nous assurer de ne pas tomber dans

le piège de ‘la collection de timbres’ comme ironisé par la fameuse boutade du physicien

10 Rutherford, suggérant que les biologistes pourraient simplement amasser des observations

sans formuler des métapatrons causals (Gallagher et Appenzeller 1999, Underwood et al.

2000, Bowen 2005). Jusqu’à présent, plusieurs études portant sur les acides gras, dans le

but de comprendre le rôle dans la nutrition et la santé des organismes, ont été menées sur

des humains ou des animaux domestiques (Dos Santos et al. 1993, Arts et al. 2001, Christie

2003, Dalsgaard et al. 2003, Speake 2004). Toutefois, les voies d’assimilation et de

circulation des acides gras des proies au prédateur sont encore méconnues, particulièrement

chez les grandes espèces sauvages telles que les cétacés. Cet état de fait est lié

principalement à la difficulté d’accéder à ces animaux, de les manipuler et donc de les

comparer aux espèces terrestres ou plus petites. Ainsi, contrairement à nos attentes

dérivées des expériences avec des organismes modèles, les quelques études effectuées

jusqu’à présent indiquent que plusieurs espèces sauvages possèdent des stratégies

physiologiques particulières pour exploiter les propriétés des différents acides gras. Les

espèces pourraient ainsi différer dans l’assimilation et la modification des lipides selon

leurs besoins métaboliques, de reproduction ou de thermorégulation (Hill et al. 1992,

Käkelä et Hyvärinen 1996, Best et al. 2003b, Pierce et al. 2004, Speake 2004, Munro et al.

2005).

L’analyses des données ou « data mining »

En plus de l’intérêt de mieux comprendre l’écologie alimentaire à travers la chimie

analytique, il existe un besoin d’élaborer des méthodes robustes d’analyse exploratoire des

données (Aitchison 1986, Andersson 2000, De'ath et Fabricius 2000, Hastie et al. 2001).

Le grand nombre de variables issu des analyses de composés peut facilement dépasser le

nombre d’échantillons. Leur utilisation comme traceurs trophiques requiert souvent

11 l’application de méthodes statistiques multivariées et non-paramétriques (Smith et al. 1997,

Dalsgaard et al. 2003, Thiemann et al. 2004a). Ces techniques exploratoires permettent

d’évaluer les patrons et les associations potentielles qui pourraient ne pas être évidents, ou

même possibles, par l’application de méthodes statistiques traditionnelles ou de modèles

linéaires (De Veaux 1995). Ainsi, les techniques d’analyse exploratoire des données sont

devenues populaires dans plusieurs domaines scientifiques, allant de la chimie

environnementale à l’écologie et à la génétique (Breiman et al. 1984, Brereton 1990,

Yunker et al. 1995, Dudoit et al. 2000).

La popularité des méthodes multivariées est en grande partie liée à l’augmentation

fulgurante de puissance et de convivialité des micro-ordinateurs et des progiciels (Afifi et

Clark 1999). Par contre, l’évolution d’outils puissants est suivie par le potentiel d’un usage

peu approprié ou incomplet pour l’analyse des données (James et McCulloch 1990, Smith

et al. 1999, Legendre et al. 2002, Beaugrand 2003, Thiemann et al. 2004a, Hobbs et

Hilborn 2006). En outre, il est souvent préférable de percevoir les méthodes multivariées

comme un moyen d’indiquer les associations potentielles, et non comme des tests rigoureux

de modèles basés sur des théories probabilistiques (Field 2000). Des mesures de puissance

d’association et la réplication des résultats peuvent être appliquées à certaines procédures.

Toutefois, la plus grande force des méthodes statistiques multivariées réside dans

l’utilisation concomitante de plusieurs types d’analyse et dans la comparaison des résultats

issus de chacune plutôt que dans l’exploitation d’une seule méthode statistique appliquée à

toutes les situations et à tous les types de données (Fig. 1.3, Table 1.2).

Jusqu’à présent, on peut déceler un biais vers certaines méthodes ou progiciels dans

la littérature scientifique. Par exemple, les analyses de fonctions discriminantes et les

12 transformations de données par ‘log ratios’ ou la mesure de distance Kullback-Leibler sont

des concepts connus chez les écologistes traitant d’analyses spatiales (McClean et al. 1998,

Burnham et Anderson 1999, Iverson et al. 2004, Silverman et al. 2004). D’autre part, la

validation statistique des composantes principales, dérivée des pratiques de contrôle de

qualité de la fabrication des produits chimiques en usine, est mieux connue par les

chimistes marins (Frank et Lanteri 1989, Eriksson et al. 2002, Grahl-Nielsen et al. 2004).

L’application d’analyses exploratoires des données ne devrait pas être vue comme une

solution pour pallier aux problèmes liés à des biais de collecte ou à des données

insuffisantes (James et McCulloch 1990, Underwood 2000). Néanmoins, ces techniques ont

le potentiel de mettre en évidence des patrons utiles à tester et à modéliser dans l’avenir, en

même temps que d’indiquer leurs limitations (Fowler-Walker et Connell 2002, Beaugrand

2003, Iverson et al. 2004).

1.2. Objectifs généraux

À la lumière des considérations sur l’écologie, la chimie des lipides et les méthodes

d’analyse de données présentées précédemment, les objectifs principaux de cette étude

étaient d’examiner les patrons inter- et intra-spécifiques des signatures en acides gras dans

l’écosystème marin du Saint-Laurent tel que révélés par :

1) l’analyse du gras du lard des mammifères marins, et particulièrement le béluga;

2) l’analyse de spécimens entiers d’une gamme d’espèces de poissons et d’invertébrés

qui représentent des proies potentielles pour les mammifères marins tels que le

béluga.

13

Au cours de cette étude, ces objectifs ont été explorés de façon plus spécifique par

l’examen de l’écologie régionale, les différences d’acides gras comme marqueurs

trophiques et l’interprétation plus élargie des différences en acides gras telles que dévoilées

par une variété de méthodes statistiques multivariées. Le corps de cette thèse est écrit sous

forme de deux chapitres indépendants pour publication ultérieure dans la littérature

scientifique primaire. Les objectifs seront présentés ainsi dans les chapitres qui suivent :

Le Chapitre 2 vise à définir les profils d’acides gras pour une banque de proies

potentielles de mammifères marins séjournant dans le secteur à l’étude pour la période

1999–2002. Les variations des profils d’acides gras des proies documentées dans ce

chapitre, tant entre les espèces qu’entre les saisons ou endroits serviront de base pour

l’évaluation quantitative ultérieure du régime alimentaire des principaux mammifères

marins prédateurs de l’estuaire et du golfe du Saint-Laurent. En ce faisant, cette banque de

données permettra également une mise à jour des informations sur les variations dans la

qualité des espèces de proie (à partir de leur teneur en lipides), ce qui permettra de mieux

comprendre la bioénergétique et le rôle trophique des prédateurs.

Le Chapitre 3 vise à comparer les profils d’acides gras de certaines espèces de

mammifères marins fréquentant l’estuaire et le golfe du Saint-Laurent: quatre espèces de

pinnipèdes et une d’odontocète (le béluga) qui sont soit des résidents ou des visiteurs de

l’aire d’étude pour la période de 1996-1998. Le béluga est d’un intérêt particulier compte

tenu de son statut d’espèce menacée dans cette région et du manque d’information sur son

régime alimentaire contemporain. À partir d’une série d’analyses multivariées, la

classification des individus basée sur leurs profils d’acides gras permettra de distinguer ou

d’associer les différents groupes. Ces comparaisons représentent un pas essentiel vers

14 l’évaluation future d’hypothèses concernant la compétition et le rôle trophique de ces

espèces et des classes de sexe et d’âge.

Enfin, le Chapitre 4 résume les résultats des analyses de profils d’acides gras des

proies et des prédateurs. Il fournit aussi des recommandations quant à l’orientation

d’études futures afin d’approfondir les résultats présentés dans cette étude.

Table 1.1. Liste des acides gras fréquemment détectés, organisés par le nom chimique,

la longueur de chaîne, et le degré d’insaturation.

Nom spécfique Nom court Nom commun

ACIDES GRAS SATURÉS tétradécanoïque 14:0 myristique pentadécanoïque 15:0 pentadécanoïque hexadécanoïque 16:0 palmitique octadécanoïque 18:0 stéarique

éicosanoïque 20:0 arachidique docosanoïque 22:0 béhénique

tétracosanoïque 24:0 lignocérique

ACIDES GRAS MONOINSATURÉS 9-tétradecenoïque 14:1n-5 myristoléique 9-hexadécenoïque 16:1n-7 palmitoléique 11-octadécenoïque 18:1n-7 vaccénique

9-octadénoïque 18:1n-9 oléique 11-éicosénoïque 20:1n-9 eicosénoïque

5,8,11-éicosatriénoïque 20:3n-9 hydromel 13-docosénoïque 22:1n-9 érucique

15-tétracosanoïque 24:1n-9 nervonique

ACIDES GRAS POLYINSATURÉS (n-3) 9,12,15-octadécatrienoïque 18:3n-3 a-linolénique

6,9,12,15-octadécatétraénoïque 18:4n-3 stearidonique 11,14,17-éicosatriénoïque 20:3n-3 éicosatriénoïque

8,11,14,17-éicosatétraénoïque 20:4n-3 éicosatétraénoïque 5,8,11,14,17-éicosapentaénoïque 20:5n-3 éicosapentaénoïque

7,10,13,16,19-docosapentaénoïque 22:5n-3 docosapentaénoïque 4,7,10,13,16,19-docosahexaénoïque 22:6n-3 docosahexaénoïque

6,9,12,15,18,21-tétracosahexaénoïque 24:6n-3 tétracosahexaénoïque

ACIDES GRAS POLYINSATURÉS (n-6) 9,12-octadécadiénoïque 18:2n-6 linoléique

6,9,12-octadécatriénoïque 18:3n-6 g-linolénique 11,14-éicosadiénoïque 20:2n-6 éicosadiénoïque

8,11,14-éicosatriénoïque 20:3n-6 homo-g-linolénique 5,8,11,14-éicosatétraénoïque 20:4n-6 arachidonique

13,16-docosadiénoïque 22:2n-6 docosadiénoïque 7,10,13,16-docosatétraénoïque 22:4n-6 docosatétraénoïque

16 Table 1.2. Sommaire des techniques statistiques multivariées fréquemment employées

lors d’analyses de données biochimiques.

Méthode Multivariée (acroynme)

Type de classification

Exemple tiré de la littérature

Analyse en composantes principales (PCA)

Apprentissage non-supervisé, ordination (Joensen et al. 2000)

Modélisation de classes (SIMCA)

Apprentissage supervisé, ordination (Marengo et al. 2001)

Régressions multiples Apprentissage non-supervisé, régression (Dalsgaard et al. 2003)

Analyses discriminantes (DFA)

Apprentissage supervisé, régression (Iverson et al. 2002)

Analyse de groupements (HCA)

Apprentissage non-supervisé, ordination (Howell et al. 2003)

Arbres de classification (CART)

Partitionnement récursif binaire

(De'ath et Fabricius 2000)

17

Figure 1.1. Schéma d’un acide gras typique, avec un carbone estérifié en tête (COO-) et

un groupe méthyle (CH3) terminal. L’exemple présenté ici concerne une

chaîne saturée (i.e., pas de lien double entre les carbones) avec le 12e

carbone en position terminale (ω).

18

Figure 1.2. Exemple de spécificité des voies de synthèse pour les acides gras à longue

chaîne polyinsaturée. Les flèches solides indiquent les voies retrouvées chez

les mammifères tandis que les flèches pointillées représentent les voies

exclusives aux autres types d’organismes. Tiré de la Fig. 1 de Leonard et al.

(2004).

19

Figure 1.3. Schémas typiques de l’analyse des données d’acides gras à l’aide des

techniques multivarieés semblables à celles utilisées au chapitre 3 : a)

démarche lors d’une analyse suivant un arbre de classification. b) démarche

lors d’une analyse impliquant une suite d’analyses en composantes

principales, de groupements, et des fonctions discriminantes. Les encadrés

gris illustrent des exemples de sorties graphiques associées avec ces

techniques de classification, soit; c) séparation en arbre de décisions, d)

projections linéaires, et e) association de cas lors d’analyses de groupements.

20

Chapitre 2

Among- and within-species variability in fatty acid

signatures of fish and invertebrate prey of the St.

Lawrence beluga whale

21

Résumé

Les profils d’acides gras de 60 espèces de proies potentielles du béluga

(Delphinapterus leucas) de l’estuaire et du golfe du Saint-Laurent ont été examinés pour

leur variabilité par espèces et autres classes. Les similaritiés et les différences révélées par

les statistiques multivariées ont suggéré que plusieurs espèces sont relativement homogènes

en composition, tandis que d’autres pouvaient varier selon la saison et l’endroit. Les

tendances générales présentées ici pourraient être utiles pour le classement des groupes de

proies tels que pélagiques ou benthiques, estuariens ou marins, et pour la modélisation

future de la diète de prédateurs. Néanmoins, la variabilité potentielle en contenu lipidique et

en composition en acides gras invite à la prudence dans l’emploi des valeurs moyennes

pour répondre à des scénarios spécifiques.

Abstract

Fatty acid signatures of 60 potential prey species of the beluga whale

(Delphinapterus leucas) of the Estuary and Gulf of St. Lawrence were examined for their

variability by species and other classes. The similarities and differences revealed through

multivariate statistics suggested that a number of species were relatively homogenous in

composition, while others could vary by season and location. The general trends shown

here might be useful for prey group classifications, such as pelagic or benthic, estuarine or

marine, et will be useful in future modelling exercises of predator diet. Nonetheless, the

potential variability in lipid content and fatty acid composition suggests caution in the

application of mean values for answers to more specific situations.

22 2.1 Introduction

The use of fatty acids to examine the diet of marine predators has interested

researchers ever since the development of tools for detailed lipid analysis made

comparisons readily possible, and it was observed that fatty acid profiles of consumers

would generally reflect those seen in their diet (Folch et al. 1957, Ackman and Eaton 1966,

Dalsgaard et al. 2003). Aquatic and marine systems, in particular, are notable for their

abundance of lipids containing longer-chained (i.e., 14 – 22 carbons) unsaturated fatty acids

that arise from plants and plankton (Beach et al. 1974, Ackman et al. 1980, Dunstan et al.

1994). By comparison, vertebrates are greatly limited in their capacities to synthesise

polyunsaturated fatty acids and must obtain these nutrients in their diet (Kanazawa et al.

1979, Tinoco 1982, Leonard et al. 2004). This limitation represents the starting point for

using fatty acid profiles as indicators of assimilated prey items in predators (Tjonneland et

al. 1993, Iverson et al. 1997a, Brown et al. 1999).

As a trophic tracer, fatty acids have already been used in a variety of ways,

depending on the subject and hypothesis being investigated (Dalsgaard et al. 2003). In

some situations, the presence of certain fatty acids in a dietary item has been used to

identify its uptake. This has been particularly the case with consumers that specialise on

classes of plants or invertebrates that are often characterised by these specific fatty acids

which otherwise may be uncommon in the local environment (Ackman and Hooper 1973,

Holland et al. 1990, Mansour et al. 2005). In order to gain use as a general tool, studies

comparing among or within species of marine vertebrates have also been conducted using

the relative abundance of a range of fatty acids, also called the fatty acid signature (Iverson

23 et al. 1997b, Budge et al. 2002, Iverson et al. 2002). Results of the signature approach,

both from captive feeding trials and from samplings of free-ranging organisms, have shown

promise for investigating trophic interactions (Kirsch et al. 1998, Phillips et al. 2003,

Walton and Pomeroy 2003, Jeffs et al. 2004). This potential for tracing diets has led to

studies attempting to relate the fatty acid signatures of predators to those of potential prey,

both qualitatively and quantitatively (Dahl et al. 2000, Kirsch et al. 2000, Olsen and Grahl-

Nielsen 2003, Andersen et al. 2004). These studies reveal the challenges in making

inferences between prey and predators using fatty acids (Smith et al. 1997, Budge and

Iverson 2003, Budge et al. 2004, Iverson et al. 2004). A number of hurdles remain before

this approach can be used with confidence on a variety of marine mammal species in the

wild (Grahl-Nielsen et al. 2004, Staniland and Pond 2004, Theimann et al. 2004). Central

to this debate is the manner in which fatty acid differences are interpreted, in particular,

how to evalutate the similarities/divergences between the signatures of assimilated prey and

those of predators.

Documenting the dynamics of fatty acid composition of potential prey is an

important step towards their use in trophic linkages (Budge et al. 2002, Iverson et al. 2002,

Phillips et al. 2002, Speake 2004). Along with species differences, fatty acid compositions

may vary with size, season or location, potentially reflecting changes in diet, but also in

response to environmental conditions or life history including population genetics

(Henderson et al. 1984, Joensen et al. 2000, Morais et al. 2003, Krahn et al. 2004). At the

same time, predators may forage over large areas, getting access to a range of potential prey

species that may vary in quantity and quality. Ostensibly, the ideal prey base would

encompass all prey types for all areas covered by a predator, although in practice, most

24 spatial diet studies should take life history into account to improve sampling design

(Iverson et al. 2004, Fritz and Hinckley 2005). The impracticalities arise because the

logistics of field sampling, along with the time and expense of chemical analyses make it

prohibitive to collect all possible variations of a prey species with replicates of 20 or more,

as might be expected for statistical evaluations using several fatty acids as variables.

Furthermore, it is not usually possible for the prey sampling to fully coincide with the

locations, seasons or years of predator sampling (Dahl et al. 2000, Hooker et al. 2001).

Therefore, examining the relative stability of the fatty acid signatures of prey, between

species and for other classes including size, season and location, is of particular interest to

validate the use of available prey data in long-term studies of marine mammal predators.

Such a need to understand prey variability is evident in the context of our long-term

study of beluga whales, Delphinapterus leucas, of the St. Lawrence Estuary, Canada (48°N,

70°W). Belugas in the Estuary form a small and relict population of approximately 1200

individuals (Gosselin et al. 2001). Although an unknown portion of the population

migrates to the northwestern Gulf of St. Lawrence during winter, many of the individuals

appear to remain in the Estuary throughout the year (reviewed in Lesage and Kingsley

1998). The population was estimated at several thousand individuals and was broader in

distribution at the end of the 19th century, but it was decimated by intensive hunting in the

first half of the 20th century (Sergeant and Hoek 1988, Kingsley 2002). Census surveys

carried out over the past two decades suggest that the population has only been increasing

at a slow rate, if at all, despite the absence of mortality from predation or hunting (Gosselin

et al. 2001).

25

The enigma of their slow recovery, and the vulnerability associated with being a

small, endemic population are matters of continuing concern for the long-term viability of

belugas in the St. Lawrence Estuary (Sergeant and Hoek 1988, Smith et al. 1990,

COSEWIC 2004). Pollution, habitat loss or degradation associated with coastal

development and marine traffic (noise, ship strikes, harassment), along with shifts in prey

abundance, availability or quality (energy density) are among the factors proposed to

explain the limited growth of this population (Sergeant and Hoek 1988, Muir et al. 1990,

Martineau et al. 1994, De Guise et al. 1995, Kingsley 2002). The lack of recent

information on beluga diet in the St. Lawrence limits our capacity to evaluate the impact of

prey sources on population health, for example when considering pathways for contaminant

accumulation or when proposing food limitation.

The only detailed information on the diet of St. Lawrence beluga comes from an

extensive study carried out in the 1930s (Vladykov 1946). This study remains the reference

for beluga feeding even though its value for understanding contemporary diets is unclear.

For example, in his study, Vladykov examined stomach contents of mostly adult males that

were hunted principally from river mouths that are no longer being used (Laurin 1982,

Reeves and Mitchell 1984, Kingsley and Reeves 1998). Secondly, changes in prey species

assemblages are believed to have occurred in the St. Lawrence marine ecosystems and

elsewhere in the Northwest Atlantic following the collapse of the commercial Atlantic cod

fishery in the 1990s (Myers et al. 1997, Sinclair and Murawski 1997, Frank et al. 2005).

The contemporary diet of beluga whales in the Estuary cannot be documented using

traditional methods such as sampling of digestive tract contents because of the precarious

status of the present population. The only beluga samples available are from beach-cast

26 strandings, of which approximately fifteen are reported annually in the St. Lawrence.

Those carcasses recovered in sufficiently good condition for a necropsy examination often

show signs of chronic poor health leading up to their death, and thus it is hardly surprising

that prey remains are very rarely reported in their stomachs. Only a few cases have been

diagnosed as trauma, i.e., individuals experiencing sudden death from ship strikes. While

these case may be good opportunities to obtain prey remains and thus information on recent

feeding events, it remains difficult to evaluate diet at the population level over long periods.

Longer-term dietary information might be obtained by examining the fatty acid

composition of blubber collected since 1988 from beluga carcasses that have stranded.

Information on diet has previously been obtained in fatty acid studies of fishes and

pinnipeds (Kirsch et al. 1998, Kirsch et al. 2000). A great deal of interest has since been

generated by the potential to identify and quantify prey contributions in predator signatures

using maximum likelihood-based approaches to modeling, such as QFASA or Quantitative

Fatty Acid Signature Analysis model developed in the laboratory of Dr. Sara J. Iverson

(Iverson et al. 2004). In order to be accurate and predictable, and therefore useful, a

number of conditions are required prior to undertaking predator diet modelling exercises

(reviewed in Iverson et al. 2004). Inferring potentially diet-linked variations in the

signatures of beluga carcasses first requires information concerning the relative stability of

prey fatty acid signatures and lipid contents, across species and over time and space. This

is necessary because belugas may be feeding in a number of different areas, from the

Estuary downstream to the northern Gulf, and the collection of carcasses span over several

seasons and years.

27

In this study, we propose to screen fatty acid compositions and lipid contents of

potential prey of beluga for the variables that are of greatest relevance for models regarding

their diet. We first consider the local ecology (prey availability) and the absolute amounts

of fatty acids (prey lipid content), before examining fatty acid compositions (prey

signatures) of potential prey of St. Lawrence beluga whales. Given the range of species and

environmental conditions that may be encountered, the characterisation of inter- and intra-

specific variability of fatty acid composition of potential prey constitutes an essential step

towards a better understanding of the contemporary feeding ecology and diet in this

endemic population of whales.

2.2. Materials and methods

Species selection

Species for the fatty acid prey base were selected using two principal sources of

information. First, we consulted local records of beluga prey, as determined from

observations and stomach contents (Préfontaine 1932, Vladykov 1935, 1946, Pippard and

Malcolm 1978). Second, we examined current prey species distributions and relative

abundance, examples of which are recorded during fisheries and scientific surveys in the

Estuary and adjacent Gulf of St. Lawrence (Therriault 1991, Bérubé and Lambert 1999,

Caron et al. 2001, Benoît et al. 2003b). As a result, approximately 60 species of fish and

invertebrates were targeted as potential prey of the beluga.

While most samples were 10–40 cm in length, smaller and larger specimens were

also sampled. Copepods (Calanus hyperboreus), euphausiids or krill (Meganyctiphanes

norvegica, Thysanoessa spp.), and hyperiid amphiphods (Themisto libellula), were included

28 to better understand the variability of fatty acid composition in the planktivorous fishes and

invertebrates that may be consumed by belugas (Dahl et al. 2000, Auel et al. 2002).

Concerning maximum prey size, there have been reports of adult male belugas consuming

specimens such as eels and sturgeon up to 80 cm in length (Vladykov 1946). However,

belugas are constrained by their relatively small oesophagus and inability to handle prey,

making them more suited to swallowing whole, relatively small-bodied prey. The other

consideration regarding maximum prey size was partially historical. Observations from

local fisheries and surveys indicate that belugas are presently less likely to encounter larger

specimens, especially of eels, salmon and groundfishes (e.g., Atlantic cod, Atlantic halibut,

Atlantic sturgeon), in the St. Lawrence compared to past decades or other areas such as

Cook Inlet, Alaska (Castonguay et al. 1994, Bérubé and Lambert 1999, Huntington 2000,

Christensen et al. 2003). Consequently, only larger specimens of salmon and eel were

available and thus included for species and size-class comparisons relative to other potential

prey.

Sample collection

The primary source for open-water samples were bottom-trawl surveys for

groundfish and shrimp conducted during the ice-free period (late April to late October) in

the Estuary and northern Gulf of St. Lawrence during 2000–2002 (Fig. 2.1). Additional

specimens of spawning and littoral species were obtained during near-shore dives and from

hand-collection by local partners at tidal fishing weirs along both the north and south shores

of the estuary. The bulk of sampling took place in 2001, with additional specimens

collected in 1999, 2000, and 2002. A number of the targeted species were encountered on

repeated occasions, either as residents or during their seasonal migrations through the

29 Estuary and Gulf. Repeated sampling for species that were at once abundant and presumed

significant prey for beluga was carried out when possible to investigate their variability

across the study region, by size-class and season. This sampling sought to compare

specimens from the Upper Estuary, the Lower Estuary, as well as the northern Gulf of St.

Lawrence, for spring, summer and autumn, between 2000-2002 (Table 2.1, Annexe II). All

samples were stored frozen at -20°C in plastic bags until processed for lipid extraction and

fatty acid analysis. Storage time before processing ranged from 6 mo. to 2 yrs.

Lipid content and fatty acid analysis

Individual specimens were thawed and measured (standard length) to the nearest 0.1

cm and weighed to the nearest 0.1 g. Samples of smaller invertebrates (zooplankton) were

comprised of pooled sets (usually 2–5) of individuals to obtain sufficient wet material for

lipid extractions. After being thawed and measured, samples were homogenised

individually in a blender or food processor. No attempt was made to remove stomach

contents prior to homogenisation, so that the specimen represented the prey item as

consumed by a predator. In some cases, additional specimens of fish that were not

processed for analysis had their stomach contents qualitatively examined for general

information regarding recent feeding events (Annexe I).

Using a modified Folch method, lipids were quantitatively recovered in duplicate

from samples of the homogenate (Budge et al. 2002). In brief, 1.5 g aliquots were extracted

with 30 ml 2:1 chloroform-methanol, and the entire lower phase was collected. This

fraction was then washed, filtered through anhydrous sodium sulphate, evaporated under

nitrogen, and vacuum sonicated to obtain total lipid mass. Fatty acid methyl esters (FAME)

30 were prepared in duplicate according Iverson et al. (1997b). The analysis of FAMEs was

conducted using a Perkin-Elmer Autosystem II capillary gas chromatograph (GC)

(Norwalk, CT) with a flame ionisation detector (FID) using a flexible fused silica column

(30 x 0.25 mm ID) coated with 50% cyanopropyl polysiloxane (0.25 µm film thickness;

J&W DB-23; Folsom, CA). Helium was used as the carrier gas and the gas line was

equipped with an oxygen scrubber. The following temperature program was used: 153°C

for 2 min, hold at 174°C for 0.2 min after ramping at 2.3°C·min-1, and hold at 220°C for 3

min after ramping at 2.5°C·min-1. Up to 66 FAME were identified according to Iverson et

al. (1997, 2001) and reported here as weight percent of total fatty acids. Each fatty acid

was described using the shorthand nomenclature of X:Yn–Z, where X represents the number

of carbon atoms, Y the number of double bonds, and Z the position of the double bond

closest to the terminal methyl group. During the analyses, several samples were identified

as containing large amounts of wax esters. These samples were subsequently reprocessed

for their fatty alcohols and then methylated to be combined with the FAMEs as derived

from the acyl lipids as per Budge and Iverson (2003), thus ensuring the quantification of all

assimilated prey lipid components.

Statistical analysis

A common requirement in statistical analyses is that the variables used be fewer in

number than the smallest sample size (Hair et al. 1998, Field 2000). This constraint poses

difficulties because of the very large number (60 or more) of fatty acids that might be

selected to serve as explanatory variables in a signature analysis (Smith et al. 1997).

Statistical analyses are therefore usually performed on a subset ( n = sample size - 1) of

fatty acids that are of greatest interest (Iverson et al. 2004). In this study, several species

31 were represented by 18 or more samples. As a result, statistical evaluation was conducted

using 17 fatty acids: 16 of the generally most abundant and diet-linked fatty acids, along

with the reference fatty acid 18:0 (often present in relatively small and stable amounts).

Prior to all tests, the data were renormalised over 100% and logratio transformed over the

fatty acid 18:0, which is the recommended data treatment for this type of compositional

data (Aitchison 1986, Budge et al. 2002).

To examine fatty acid compositions while including those species only available in

small sample sizes, one solution is to conduct a Principal Component Analysis (PCA). This

approach summarises the original set of variables into a smaller set of independent factors.

Although PCA is not strictly limited in the number of variables to be used, measures of

adequacy and the interpretation of results are usually improved with a restraint on the

number of input variables and the use of appropriate data transformations (Bonn 1998,

Osborne and Costello 2004). The PC analyses were therefore restricted to using the set of

16 logratio fatty acids. From the analysis, only those PC factors with eigenvalues >1 were

retained as significant. Following a Varimax rotation of axes (Kaiser 1958), the greatest

absolute loadings of the fatty acid variables contributing to each retained factor were

evaluated. The results were also summarised as two-way plots by factor for the individual

scores.

Establishing predictive rules with which to distinguish samples into known classes

while using several variables is a typical application of Discriminant Function Analysis, or

DFA (Bradshaw et al. 2003). For this investigation, DFA was used in a more exploratory

fashion, to reveal the extent to which defined groups may be readily distinguished using the

set of fatty acids (Field 2000). To achieve this goal, the classification results were visually

32 summarised by the first two significant discriminant functions, or canonical variates, and

presented in biplots of the fatty acid variables along with individual scores and their 95%

confidence intervals over group means.

Variability in lipid contents between samples was examined using one-way

ANOVAs followed by a pairwise comparison of means using post hoc Tukey’s test.

Significance was established throughout at a = 0.05 All statistical analyses were performed

using the JMP 5.1© (SAS Institute Inc, Cary, NC) software package.

2.3. Results

Representation in prey base

A total of 1028 samples were analysed for their lipid content and fatty acid

composition, with major taxa including five species of zooplankton, 14 species of

crustaceans (shrimps and crabs) and molluscs (cephalopods and gastropods), and 40 species

of demersal and pelagic fishes (Table 2.1). Among the repeated samples were three pelagic

(herring, capelin, and smelt) and four demersal (plaice, flounder, tomcod and shrimps)

species. Other potentially important prey of beluga, at least historically, failed to be

obtained for all desired periods or location, as in the case for Atlantic cod and sand lance in

autumn, or Arctic cod and shortfin squid that were encountered only in the Gulf.

Lipid content

Mean amounts of total lipids in percent wet mass ranged from < 1% to over 20%

among species groups (Table 2.2). In general, lipid contents were highest (over 12%) in

copepods and anguilliform fish (eels, lampreys, and hagfish), followed by the large pelagic

fish species (salmon, shad, herring, barracudina, and Greenland halibut) with mean amounts

33 of 6–12% lipids, the smaller pelagic species (sand lance, capelin, shortfin squid, smelt,

daubed shannies, euphausiids and amphipods) with 3–6.5% lipids, and the decapod

crustaceans (shrimps and crabs) and most of the demersal fish species (sculpins, flatfishes,

and codfishes) with mean lipid contents of 2% or less. Exceptions were longfin hake and

pollock that occasionally had lipid contents several times those of other codfishes.

As expected, season significantly affected lipid contents in a number of groups, but

these variations differed with size class and sampling location. For example, lipid contents

were generally lower in the spring compared to summer or autumn in large herring (>18

cm) and capelin whereas lipid contents of smaller herring (<18 cm) did not change

significantly between spring and autumn. The lipid content of smelt also varied

significantly with season, but patterns varied regionally. Smelt from the Upper Estuary

differed significantly in their lipid contents between May (1.6%) and October (3%), while

smelt from the Lower Estuary showed high lipid contents in July (6%) and October (6.6%)

and lower levels in intervening periods. In contrast to the pelagic species, lipid contents of

shrimps and most demersal fishes were less affected by seasonal changes, but these latter

groups often had lower absolute amounts of lipids. For example, lipid levels in similarly

sized tomcod rose from 2% in spring to 3.5% in late summer. For plaice, individuals were

only marginally richer in lipid in spring compared to summer or autumn (means = 2.2% vs.

1.4% and 1.5%, respectively, one-way ANOVA F2,39 = 3.3, P =0.049).

General fatty acid compositions

The analysis identified up to 67 fatty acids and their isomers, 25 of which had global

mean amounts >0.4% each (Table 2.2). The subset of 17 fatty acids retained for statistical

comparisons were 14:0, 16:0, 16:1n-7, 16:4n-1, 18:1n-9, 18:1n-7, 18:2n-6, 18:3n-3, 18:4n-

34 3, 20:1n-11, 20:1n-9, 20:4n-6, 20:5n-3, 22:1n-11, 22:1n-9, 22:6n-3, and the reference fatty

acid 18:0, which together accounted for an average 88% of the total fatty acids by mass.

The selection of abundant fatty acids or those known to be associated with diet is

one approach towards facilitating the identification of dietary components. However, the

use of a restricted set (17 vs. 25 or 67) of fatty acids can result in the loss of useful

information for this task. Several fatty acids were detected in significant amounts in certain

species, but were in very low levels when averaged across all species. This was the case for

the fatty acid 22:1n-7 (global mean = 0.27%), which averaged >0.4% in daubed shannies,

plaice, herring, and several species of crustaceans (Fig. 2.2). Similarly, the fatty acid

14:1n-9, was found in amounts > 0.5% in gastropods including whelks, and in flatfishes

such as witch flounder, Greenland halibut, and plaice. However, as neither of these

monounsaturated fatty acids is well understood with regards to dietary origins, they were

not considered further in comparisons. Several other fatty acids were detected at similarly

low levels (i.e., ≤ 0.5%), but were retained because of their better-known association with

dietary sources. This was the case for fatty acid 18:2n-6, 18:3n-3 and 18:4n-3 that were

found in substantial levels only in estuarine species such as salmon smolts, eels, nereid

worms, snailfishes, and smelt (Fig. 2.3). Similarly, the fatty acid 16:4n-1 was generally

detected in trace amounts with the notable exception of herbivorous zooplankton including

Thysanoessa spp. (krill) and Calanus hyperboreus (copepod), but also certain fishes,

notably herring sardines, sand lance, and smelt.

Other polyunsaturated and monounsaturated fatty acids were detected globally in

substantial amounts, and exhibited wide inter-specific variability (Table 2.2). The fatty

acid 22:6n-3 was especially abundant in marine demersal fishes (hake, cod, rockling,

35 haddock) and in squid, octopus, lumpfish, and skates, although each of these also had a low

fat content and therefore relatively low absolute amounts of fatty acids. Similarly, the

polyunsaturated fatty acid 20:5n-3 was often abundant in several species that had low total

lipid contents, such as crabs, shrimps, whelks, and octopus. Another marine-linked fatty

acid, the monosaturate 20:1n-9, was seen in greatest relative abundance in copepods,

followed by hyperiid amphipods, the larger specimens of sand lance, herring, and shad

(each averaging >15% for this fatty acid). In contrast, eels, salmon smolts, and shad

sardines had very low amounts of this fatty acid, averaging 1% or less.

Multivariate comparisons among and within species

An exploratory PC analysis examined all specimens, both the species in low

numbers alongside the more-abundant groups. This analysis reduced the set of 16 fatty acid

logratios into four significant factors that together accounted for 82% of the original

variance (Table 2.3). Following Varimax rotation, Factors 1, 2, 3, and 4 explained 18, 25,

18, and 21% of the variance, respectively. Correlations of fatty acids 20:5n-3 and 22:6n-3

were negative with Factor 1. Fatty acids 18:2n-6 and 18:3n-3 were strongly positively

(>0.8) on Factor 2, along with high loadings for 16:0, 16:1n-7, 18:1-9 and 18:1n-7. Factor

3 was strongly negative for 20:4n-6 (>0.9), but positive for 14:0, 16:5n-3, 18:4n-3. Factor 4

was characterised by positive loadings for the 20-series mononunsaturated fatty acids of

20:1n-9, 20:1n-11, 22:1n-9, and 22:1n-11. A two-way matrix plot of the rotated PC factors

revealed that most samples had relatively similar scores, although some species were

segregated (Fig. 2.3). Species that were projected farthest from the origin of axes included

estuarine fishes (salmon smolts, shad sardines, and eels) and smaller crustaceans (eualid

shrimps, amphipods, and copepods).

36

We next carried out a series of pairwise comparisons of the retained factors, aimed

at elucidating patterns among selected sets of samples. Thus, in a new PCA based only on

eight prey species commonly found in the Estuary in autumn, two PC factors were

sufficient to summarise 80% of the total variance. Examining the rotated factor loadings,

Factor 1 (51% of variance) was characterised by a negative correlation with 20:4n-6 and a

strongly positive correlation with the marine-linked fatty acids 16:4n-1, 18:4n-3, 20:1-9,

20:5n-3, 22:1n-9, 22:1n-11, and 22:6n-3. Factor 2 (29% of variance) was most strongly

correlated with the polyunsaturated fatty acids 18:2n-6 and 18:3n-3, but also with the

common monounsaturated fatty acids 16:1n-7, 18:1n-7, and 18:1n-9 (Table 2.4). Plotting

sample scores on these two factors had eels segregated from pelagic species on the basis of

low values on Factor 1 and from demersal species on high values on Factor 2, with smelts

interspersed among the latter two groupings (Fig. 2.5a). While seeking to enhance the

segregation of species in the two-way plots, it was observed that groups with positive

scores on Factor 2 also demonstrated relatively high lipid contents (Fig. 2.5b). A

reevalution of of the two-way plot in the PCA using all specimens also indicated that the

additional character of lipid content could serve along with the PC Factors to highlight

certain species group, such as eels that scored positive on Factor 1, and copepods on Factor

3, while hagfish had positive scores on both of these factors (not shown; Annexe III b).

As a separate PCA for autumn samples was similar in projection relative to the the

analysis when using all specimens (not shown), an examination of selected species scores

was continued based on the results of the original PCA. In one example, the factor scores

of four species of the cod family were compared because they represented a diverse and

ecologically-important group. Partial segregation between Atlantic cod and several tomcod

37 was suggested on Factor 1. While there was considerable overlap in scores between the two

species, the majority of tomcod were positive on Factor 1 and revealed smaller

contributions of 20-series polyunsaturated fatty acids compared to Atlantic cod (Fig 2.6).

Also seen with Factor 1 was a possible separation between pollock, with the three

specimens from the Gulf scoring closer to Atlantic cod than were three specimens from the

Estuary. Positive scores on Factor 3 appeared to segregrate pollock and Arctic cod from

Atlantic cod. Another potential distinction was seen with Factor 4, due to greater amounts

of the 20-series mononunsaturated fatty acids in many of the larger Atlantic cod that scored

positive as compared to Arctic cod and most tomcod. When combined, Factor 1 and 4

appeared to be the best segregators of Atlantic cod from the other gadid specimens in this

total PCA series.

The factor score plots for the pelagic group of capelin, smelt, and herring were next

explored in order to evaluate the limits of the information that may be obtained from these

similar species when using PCA. A plot restricted to the first two PC Factors for the three

species confirmed their broad overlap in scores, principally on the basis of all

polyunsaturated fatty acids, with the exception of 20:4n-6 (Fig. 2.7). A full matrix plot

suggested that Factor 3 could be used to segregate most capelin and herring from smelt,

largely on the basis of the greater amounts in fatty acid 20:4n-6, while Factor 4 generally

distinguished capelin from herring, due to greater amounts of 20-series monounsaturated

fatty acids in the latter species. Among these repeated samples, capelin varied the least on

the first two Factors, as sugggested by their smaller distribution of scores (grey-filled

convex area) relative to herring or smelt (Fig. 2.7). Detailed examination of the species

revealed that capelin, on the whole, were least variable on Factor 1 (grouping near -1), but

38 some separation between samples, according to location, season and year, was apparent

along Factor 2 (Fig. 2.8; see full matrix, Annexe III c). In contrast, when examining all

three species for one subset of samples (autumn in the Upper Estuary in 2001), once again

no single PC Factor clearly separated the groups, except that sardines (small herring) loaded

negative relative to large herring (18 cm or more) on Factor 1 (Fig. 2.9).

As seen earlier, when presented with similarity in factor scores, or species overlap,

an examination of lipid content occasionally served to enhance group segregation, though

not necessarily by species. For example, some of the fattier smelt were noticeable as

scoring higher (>1) on Factor 2 than other smelt and the capelin or herring, while the fattest

herring (>10% lipid) more closely resembled capelin on both Factors than they did to either

the leaner herring or the smaller (and also leaner) herring sardines (Fig. 2.10).

Following these general explorations using PCA and total lipids, differences in fatty

acid signatures were next compared using a series of discriminant function analyses. An

initial analysis was restricted to seven of the most-abundantly sampled species groups. In

this scenario, plotting on the first discriminant function had capelin scoring closest to

herring, while tomcod, crangon shrimp, and flatfish appeared closer to the origin (less

negative scores). Smelt were distinguished from all other species by their positive scores on

the second discriminant function. Despite some overlap in individual scores (not shown),

each of the species group means were significantly segregated, as suggested both by the

separation of their 95% confidence interval ellipses of group means and by the very low

number of individual misclassifications 4%, or 20 of 499), half of which were between

capelan and herring (Fig 2.11).

39

A subsequent discriminant analysis was conducted on all groups with sample sizes

greater than 17, thereby making full use of the logratio set of fatty acids. All smaller

groups were also plotted on the resulting discriminant functions in order to suggest

potential associations for future exploration with larger sample sizes. For the groups used

in the analysis, the misclassification rate was 10% (70 of 699). In plotting individual scores

on Discriminant Functions 1 and 2, the whelks were strongly segregated from all other

species, principally on the basis of their lower amounts of fatty acids 16:0 and 16:1n-7.

Eels were the next most distinctive species from the remaining groups, essentially on the

basis of very low amounts of several monounsaturated fatty acids.

A subsequent analysis that excluded whelks and eels showed more clearly-

discernible groupings of demersal fishes, pelagic planktivores (fishes and

macrozooplankton), and decapod crustaceans (Fig. 2.12). The first discriminant function

appeared to separate marine and pelagic from estuarine and demersal species, while the

second discriminant function mostly segregated crustaceans from fishes. Those samples

grouped in the scores plot as ‘marine’ or ‘pelagic’ were generally associated with an

abundance in ‘plankton-linked’ fatty acids, 16:0 and 22:1n-11, and lower amounts of ‘plant

detritus-linked’ 20:4n-6 as compared with those samples that scored towards the ‘estuarine’

or ‘benthic’ labels. The decapod shrimps were associated with high levels of fatty acids

16:1n-7 and 18:1n-7. Smelt continued to be intermediate between the other two putative

fish groupings on the first discriminant function, and oriented towards eels when examined

on the second discriminant functions. Of interest among the individual classifications, four

Greenland halibut, obtained near the Upper Estuary, scored closer to the estuarine

demersals compared to other specimens that appeared in the group of pelagics.

40

In light of these results, further discriminant analyses tested the efficacy of group

classifications as suggested from the species analysis. For example, spring of 2001 was a

period when a number of of both pelagic and demersal species were obtained at a time

when the spring spawning of capelin is the focus of activity in the local marine ecosystem.

A discriminant analysis for this period was highly successful (91%, or 95 of 102 correctly

classified) at placing specimens into one ot two functional groups as either pelagic or

demersal primarily on the basis of fatty acids 14:0, 18:4n-3 and 22:6n-3. Of note, five of

the nine misclassified specimens were of larger specimens of tomcod (18–32 cm). These

individuals, along with two flounder and one crangon shrimp, were predicted as belonging

to the pelagic group that included capelin, principally on the basis of their higher amounts

of 22:6n-3 (Annexe III d).

Autumn was another period chosen to examine group patterns in signatures, to

evaluate potential differences after a summer of feeding. A discriminant analysis

classifying species from autumn samples (2000 and 2001) highlighted the segregation of

eels on the basis of their low amounts of 20:1n-9, 22:1n-11 and 22:6n-3 (not shown). Once

again, excluding the distinctive species (eel) and forming composite species groups

revealed four broad, but still-significant, groupings. In this successful analysis (0 of 156

misclassified), the pelagic (smelt, herring, and capelin), demersal (flatfishes) and marine

shrimp (eualid and striped shrimps) groups scored apart from the crangon shrimp group,

principally on the basis of the major polyunsaturated and 18-series mononunsaturated fatty

acids that contributed to the first discriminant function. On the second discriminant

function, with major contributions from 20-series monounsaturated fatty acids, pelagic

41 fishes were segregated from flatfishes, marine shrimp in one direction and crangon shrimp

in the other (Annexe III e).

Digestive tract contents

Most fish species in this exploratory study had small sample sizes and consequently

all available specimens were processed for the chemical analyses. However, some sample

groups, especially of smelt and tomcod, were obtained in greater numbers than was needed.

The dissection of digestive tracts in these extra specimens was an opportunity for potential

insights into regional and seasonal feeding activities of a limited number of groups in

parallel to the lipid/fatty acid analyses (Annexe I). The dissections were subjective and

selective, and the small sample sizes were not intended to quantify stomach contents, as

general diets may be found in the literature for these common species. The purpose was to

examine members within a group (by species, location, or period) for ‘significant’ lipid-rich

prey remains that could be assumed to contribute to the fatty acid signature of a similar

specimen when homogenized whole, and in turn, to the variability of the signature for that

group. This bias in using whole specimens is usually assumed rather than being examined,

and therefore even fragmentary observations may be helpful in suggesting recent trophic

ecology, especially for a particular region and period of a species group.

For the Upper Estuary in May, several of the larger (>17 cm) tomcod had stomachs

with capelan (4 of 16 specimens) and crangon shrimps (7 of 17). Also seen for spring

samples were copepods in 9 of 23 tomcod (14–22 cm), 3 of 4 smelt, and 2 of 2 sand lance.

Krill filled the stomachs and intestines of examined barracudinas from the Lower Estuary,

with mostly Thysanoessa sp. for 5 individuals in May, while Meganyctiphanes norvegica

42 were seen in 3 individuals for July. Krill (non-specific) were also seen in 1 of 2 specimens

each of snailfish and Atlantic poacher from the Lower Estuary in July, and in 1 of 9 smelt

from the Upper Estuary in August. While 12 of 27 smelt examined were empty, all tomcod

specimens contained some material in their stomachs, even if the prey items were not

always recognizable. Both smelt and tomcod were similar in containing traces of

polychaete worms and amphipods in a number of specimens from spring through autumn.

In contrast, traces of insects were only seen in 3 of 60 specimens of tomcod and smelt.

Shad sardines were the only other group examined that had insect remains, with flies and

larvae in 2 of 4 specimens that contained prey items.

2.4. Discussion

The comparative analysis of the lipid biochemical fraction of potential prey of St.

Lawrence beluga revealed both intra- and inter-specific patterns in lipid contents and fatty

acid compositions. While this study illustrated the possibility to distinguish prey species on

the basis of fatty acid composition, it also suggests that success at classification will be

enhanced, both at the individual level and for broader classes, by making use of

complementary information and methods. Examples of complements demonstrated here

include accounting for different species origins or trophic niches, total lipid quantity,

successive multivariate analyses on subsets of data, and digestive tract contents. Several of

these elements may be routinely applied in diet-based and fatty acid studies, but less

common is the attempt to make simultaneous use of all this information to understand

species and spatio-temporal patterns in fatty acid signatures. It is with this frame in mined

that the study results and the choice of examples in each section are discussed in detail

below.

43

Species sampling

The first step towards optimising fatty acid analysis for diet modeling is to establish

the availability of potential prey, as rare or less-preferred species are unlikely to be detected

in the predator signature (Iverson et al. 2004). Failure to detect infrequent dietary items is

also a concern for techniques with specific identification including stomach contents and

fecal DNA (Deagle et al. 2005, Svenning et al. 2005). This constraint makes it difficult at

times to establish evidence of significant predation on threatened stocks of commercial

species including cod, salmon and eels, which is a frequent objective of diet studies

involving marine mammals and seabirds (Vladykov 1981, Shelton et al. 1997, Orr et al.

2004, Svenning et al. 2005). Another example where large prey species information might

be desired but unavailable is the identification of pathways of environmental contamination

in predators, such as linking eels with belugas (Hickie et al. 2000). At the same time, this

limitation when using fatty acid signatures can be helpful, by focusing attention on known-

significant prey, i.e., because of abundance or preference, as being most likely responsible

for the patterns detected (Andersen et al. 2004, Iverson et al. 2004). In the case of eels, the

consumption of this potential prey by belugas might fall into doubt, given the marked

decline in recent years of eel populations (Castonguay et al. 1994, Savaria et al. 2003).

Along with relative abundances, environmental changes are suspected to have led

the St. Lawrence to harbour different species assemblages than was observed in past

decades (Hanson and Chouinard 2002, Benoît et al. 2003a, deYoung et al. 2004). In this

study, two pelagic species, shortfin squid and Arctic cod, have been reported as prey of

beluga either historically in the St. Lawrence or elsewhere in the Arctic. During sampling

44 from 2000-2002, both of these schooling species were only encountered in the the Gulf.

Although it is possible that these species were present in the Estuary, it is suggested that

only the occasional migration of individuals would have likely have experienced significant

access to these and other common Gulf species, such as larger members of the cod family.

This regional difference in potential prey distribution may be an example of species

association with alternating environmental conditions (Dawe et al. 2000, Colbourne et al.

2002). For example, squid embark on feeding migrations, coming into the Gulf from the

South (Nova Scotia) in late summer, occasionally being reported as far as the Estuary,

presumably in years with warmer water temperatures (Préfontaine 1930, Mercer 1970).

Conversely, colder-water species such as Arctic cod and its principal prey the hyperiid

amphipod, Themisto libellula, descend into the Gulf from the North (Labrador) during cold

events, as attested to by marine survey reports and predator stomach contents in the years

leading up to the the sampling period of 2000-2002 (Dempson et al. 2001, Harvey et al.

2004). The Northwest Atlantic on the whole was associated with a negative condition

index during this period, even though water temperatures in the Gulf increased relative to

the 1990s (deYoung et al. 2004, Drinkwater and Gilbert 2004, Gilbert et al. 2004, Harrison

et al. 2004).

This appeal in correlating observations of a species with an environmental condition

index is thus often tempered by conflicting information when examined from several

different indices, in both scale and type, making it a challenge to draw conclusions

regarding impacts and causes of changes in prey species, and in turn, to predators

(Hjermann et al. 2004). Notable other examples of linkages that have been attempted

include the role of water temperature with capelin and seabirds (Carscadden 2005, Davoren

and Montevecchi 2005), atmospheric depressions (i.e., NAO) with copepods and right

45 whales (Greene and Pershing 2004), and Alaskan water circulation with walleye pollock

and Steller sea lions as reviewed in Fritz and Hinckley (2005)

Environmental changes aside, sampling success was not necessarily a reliable

indication of the availability of prey in the study region. For example, autumn-winter is an

important feeding period for several marine mammal species, possibly including the beluga

(Lydersen et al. 1989, Martin and Smith 1999, Winship et al. 2002, Laidre and Heide-

Jørgensen 2005). Unfortunately, limited sampling opportunities in autumn made it difficult

to evaluate several prey species for spatiotemporal differences, in particular Atlantic cod,

capelin, and sand lance. Thus, cod may have been unavailable in autumn because of low

abundance or insufficient fishing effort in the estuary. Eels were available from the autumn

fishery, but this brief appearance (usually a few weeks in October) arises because of their

reproductive migration to the sea, and it represents the only opportunity for predation

during the year. Capelin were not commonly encountered in summer and rarely captured in

autumn, although harp seals are known to consume large amounts in winter (Beck et al.

1993, Stenson et al. 1997) and this fish is ubiquitous during the spring spawning period.

The seasonal differences in availability is probably an indication of the unreliability

of sampling pelagic species when using groundfish and shrimp nets, rather than the absence

of these species in the ecosystem. Such a difficulty was most apparent by the virtual

absence of sand lance during fish surveys except when using smaller dredges and sampling

gear (pers. obs.). Sand lance may represent a significant prey species of belugas in autumn

(Vladykov 1981), as also revealed by digestive contents in a sturgeon, seals, and a minke

whale from the Lower Estuary during this period (pers. obs.). Other species, both resident

(tomcod, shrimps and flatfishes) and migrant (herring) were both available and abundant in

46 all seasons from inshore fishing activities, which may be valid indications of their potential

as important prey sources and for evaluation by standard fishing practices,

Lipid contents

Along with regional and seasonal availability, the lipid content of prey species may

play a significant role in the diet of a marine predator, which may explain the attention

given to this trait among prey species when examining declines of predator populations,

notably of seabirds and pinnipeds (Kitts et al. 2004, Abookire and Piatt 2005, Rosen and

Trites 2005, Wanless et al. 2005). Prey preferences are often assumed to be based on

energetic values as indicated by the lipid content (Lawson et al. 1998b, Lea et al. 2002b).

If all else is equal in foraging effort, a predator may be expected to seek higher-lipid prey

items. Furthermore, while changes in fatty acid signatures can be detected even when

feeding on a low-fat diet, e.g., Kirsch et al. (2000), the prey signature is usually assumed to

be more readily apparent (i.e., faster detection of significant changes to predator

composition) when consuming fatty specimens relative to leaner prey (Santos and Pierce

2003, Iverson et al. 2004).

From the samples obtained in this study, the lipid contents of flatfishes and decapod

crustaceans varied little with season, location, or size-class, thereby suggesting that the

mean values presented for these groups will not likely be improved by additional sampling.

Other species demonstrated significant variability, thereby highlighting the need to confirm

patterns such as the generally increased lipid contents seen in larger fish specimens (>18

cm) and in later seasons (spring vs. summer and autumn). Within-species differences in

lipid contents were the most pronounced for herring, capelin and smelt. Herring were

47 relatively systematic, with predictable differences being seen between seasons and size-

classes, as compared to smelt and capelan which could reveal greater variability between

months and locations. These forage species were obtained in repeated samplings, but under

varying conditions (on shore or at sea) while they were spawning, feeding, or migrating

during the sampling period. The range of somatic and environmental conditions

encountered by these forage species makes it pertinent to examine their variance in addition

to their group (i.e., by size, season, or location) means in order to accurately capture the

lipid dynamics of interest. For example, documenting temporal variability in lipids of

capelin may point to the impact on the fitness of seabirds that depend on this prey species

(Davoren and Montevecchi 2003).

Unlike most pelagic fishes, codfishes are typically characterised as lean, with about

2% lipid by wet mass. Of interest was the relatively high amounts of lipid seen in pollock

which is very similar morphologically and overlaps in distribution with the Atlantic cod.

The pollock from autumn revealed the greatest amounts, averaging >5% lipid, while other

specimens of pollock and other codfishes were all sampled earlier in the year. In the

Northeast Pacific, walleye pollock (Theragra chalcogramma) increased in lipid content

from spring to autumn with levels comparable to the pollock sampled here (Kitts et al.

2004). However, pollock are not caught often in the Estuary and the Gulf (pers. obs.), and

are more commonly associated with banks in the Northwest Atlantic (Scott and Scott 1988,

Benoît et al. 2003b). Thus, in parallel with the previous example given with eels, the

potential prey value from the higher lipid content in pollock is likely to be discounted

because of their less-favorable distribution for predation by St. Lawrence belugas.

48 General fatty acid comparisons

This study indicated that the greatest source of variability in a selection of fatty

acids from the collected prey species were polyunsaturated fatty acids, and to a lesser

extent, mononunsaturated fatty acids. While several of the fatty acids may originate

through diet, the levels of specific fatty acids were not always easily attributed to

differences in diet. Thus, a number of the fatty acids may be used as diet-linked tracers at

certain trophic levels, e.g., 16:0, 16:1n-7, and 16:4n-1 in plankton, but were less likely to be

attributed to their direct prey origins, for example, the consumption of diatoms, when

assimilated into top predators. Mammals commonly modify a number of saturated and

mononunsaturated fatty acids (i.e., 14:0, 16:0, 16:1n-7, 18:1n-7, and 18:1n-9) or contain

only trace-level amounts in their adipose tissues of polyunsaturated fatty acids, as seen with

fatty acid 16:4n-1 in this and other studies (Dalsgaard et al. 2003, Iverson et al. 2004).

When associated with origins in certain prey species, fatty acids of dietary interest

may sometimes be characteristic of their environment. In a common example, 18:2n-6 is

abundant in terrestrial plants and freshwater microorganisms (Desvilettes et al. 1997),

which lead to food sources for estuarine fishes that includes eels, smelt, shad sardines and

salmon smolts. By comparison, most species examined in this study revealed an abundance

of marine-linked fatty acids, suggesting the common base of phytoplankton with fatty acids

20:5n-3 and 22:6n-3 (Sargent et al. 1985, Fraser et al. 1989) and copepods with fatty acids

20:1-9 and 22:1n-11 (Kattner and Krause 1987) in supporting a variety of species at higher

trophic levels. Examples seen in this study were the relative amounts detected for cod and

crabs (20:5n-3, 22:6n-3), and sand lance and herring (20:1n-9, 22:1n-11). Conversely,

certain polyunsaturated fatty acids, such as 20:4n-6 and 22:5n-3, may be present in

49 vertebrate tissues in major quantities, but are thought to reflect metabolic rather than trophic

activities (Imfeld 2000, Ferdinandusse et al. 2001). Occasionally abundant in invertebrates

such as crabs and whelks, the fatty acid 20:4n-6 is found in relatively-low amounts in

storage fats of mammals, perhaps due to its essential role in membrane function rather than

energy storage (Cooper et al. 2005). Conversely, the fatty acid 22:5n-3 may be detected in

relative abundance in higher vertebrates, but it is viewed as a metabolic intermediate

between 20:5n-3 and 22:6n-3, thereby lessening its value as a dietary tracer if compensation

is not performed (Pawlosky et al. 2003, Iverson et al. 2004).

Inter- and intra-specific fatty acid variability

Comparisons using PCA and discriminant analysis on a diet-linked subset of fatty

acids provided insights into general and specific patterns, both when examining the entire

prey base, and when evaluating selected groups between and among species. A PCA using

all prey samples had their variance in fatty acid composition summarised into four

significant factors, confirming the segregation of estuarine fishes such as eels, along with

salmon smolts and shad sardines from most other species which scored as being relatively

indistinct. The similarities among most of the remaining species may be a broad reflection

of mixed, but mainly marine-linked (i.e., marine plankton-based) diets, especially for

abundant forage species like herring, capelin, and smelt, followed by piscivores such as

larger codfishes and flatfishes.

Broad similarities in diets among the groups shown here might be expected to result

in approximately similar fatty acid profiles, particularily given the emphasis on evaluating

the generally abundant and diet-related fatty acids rather than potentially-discriminating

traces. The suggestion of shared diets was also indicated partially by the stomach contents

50 of extra specimens that were occasionally available. For example, a number of specimens

contained krill and copepods in several species for all locations, especially for the period of

May through July. Lipid analyses that seek to relate prey contributions to a predator’s diet-

linked fatty acid signatures are best conducted using whole, homogenized prey items in

order to represent actual diets, as opposed to removing digestive tracts or analysing selected

tissues such as muscle (as is often done with squid and fish). In the case of belugas, it may

of be particular interest to examine further the intermediate prey remains in their direct

prey. An example is the copepod- and krill-related signatures that may be supplemented in

fishes such as sand lance and barracudina that ingest these items, especially in spring and

summer. A fatty acid food web study in Svalbard suggested that the copepod-like signature

in sampled belugas in Svalbard was an indication of consumption of arctic cod, which in

turn preyed on copepods, rather than to the direct ingestion of copepods by belugas

themselves (Dahl et al. 2000). Thus knowledge of local food webs as revealed by

examining intermediate prey may help in interpreting the path of lipids revealed in analyses

at higher trophic levels.

The examination of specimens belonging to the same set as those analysed for fatty

acids thus represents an opportunity for partial information on fatty acid signature

variability as arising from digestive tract contents. For example, a large fish such as cod or

turbot may contain sufficient shrimp biomass in their stomachs such that a whole-specimen

analysis is affected by recent feeding on shrimps, in addition to the past dietary history of

the predator as stored in its body lipids.

Along with identifying idividual constituents in terms of fatty acid, the analyses also

provided information on total lipid content of samples. Segregation of potential prey

groups for predator diet models might be assisted by considering lipid content as a

51 supplementary variable when statistical analyses produce broad or indistinct groupings.

Such a scenario might be seen with the PCA results. In this case, the characteristic of lipid

content could enhance the differences among groups from that seen when plotting

individual factor scores derived from the set of fatty acids. An example was in the analysis

for autumn samples, where several species groups that had similar factor scores on the

retained PCs were further distinguishable when considering their lipid contents, specifically

between individual smelt, between smelt and herring, and between large and small herring.

Overall, PCA was most useful at characterising the fatty acid variables (as

compared to cases, or species) by summarising the selected set of fatty acids into functional

groups. Thus, PC Factors 1 and 4 were most strongly-associated with the marine plankton-

linked fatty acids, 20:5n-3 (diatoms) and 20:1n-9 (copepods), respectively. PC Factor 2

was defined more by the minor polyunsaturated fatty acids 18:2n-6 and 18:3n-3, likely to

be associated with freshwater species (Desvilettes et al. 1997, Budge 1999). Contributions

to Factor 3, while not as strong, could still be linked to diet when considering the plankton-

linked minor fatty acids of 16:4n-1 and 18:4n-3 (Budge 1999). Factor 3 also exhibited a

strong negative correlation with 20:4n-6, making it likely an indicator of benthic detrivory,

as may have been suggested by the scores of species such as gammarid amphipods and

eualid shrimps.

While the factors produced through the PCA could be associated with functional

groups of species, understanding the variability within species was less straightforward.

Hence, even when subsequent PC analyses were restricted to selected groups, such as

samples from a particular season (autumn) or a family (codfishes), few groups could be

strongly differentiated in the two-way factor score plots. This is a concern because

inferences will be less persuasive when based on overlapping groups. At the same time, the

52 lack of differences suggests a large deal of commonality, as might be expected in food webs

with sympatric groups of species (Dahl et al. 2000, Lesage et al. 2001, Andersen et al.

2004). In summary, PCA was useful because it enabled the inclusion of information from

smaller sample sizes, which was the case for the majority of the 60 species analysed here.

However, it also appeared to lose desired information for tracing diets, as groupings among

and within species were largely not differentiated when using this data-reduction technique.

Discriminant analysis had greater success at segregating samples, as was shown by

the significant separation of group means projected in score plots and relatively low

misclassifications rates (usually <10%). The relative success, however, was accompanied

by the requirement for sufficient samples in order to make use of a large number of

classification variables. With this in mind, of additional interest was the graphical use of

the classification results to reveal novel patterns, both in logical classes such as species and

season, but also for trophic-linked, functional groups, i.e., pelagic vs. demersal species.

Individuals of species or other categories can be closely associated, as indicated by the

apparent clustering of group means along the discriminant functions. Thus, even when

known classes can be significantly-discriminated when using large numbers of samples and

variables, it may be useful for subsequent tests if categories could be established that were

readily distinct as well as interpretable for diet sources. In the series of discriminant

analyses performed here, groups were readily characterised as pelagic vs. demersal, marine

vs. estuarine, and fish vs. invertebrate, based on the proximity of specimen scores on the

first two discriminant functions, along with the proportional contributions of the fatty acid

variables and presumed life-history of species (Scott and Scott 1988, Squires 1996, Bourget

1997, Nozères et al. 2003)

53 Obtaining knowledge of potential enclosing groups with associated variables is

useful for two reasons. First, it may allow diet determinations when specific identification

of prey sources is less feasible due to similarities in fatty acid composition or insufficient

sample sizes for full testing. New data for other prey species, or even for predators, could

then be projected on the derived functions, enabling their comparison with the specified

groups, to test the identity of proposed relationships. Such tests have commonly sought to

establish differences between predators’ diets based on fishes vs invertebrates (molluscs or

crustaceans) (Bradshaw et al. 2003, Phillips et al. 2003). In one example seen in this study,

whelks and eels were projected into distinct groups of benthic invertebrates and estuarine

fishes, respectively, based on discriminant functions calculated on the more abundant

samples (Fig. 2.12). Two other groups, hyperiid amphipods and shortfin squid (plotted, but

not labelled in the figure because of small sample sizes) scored nearer to smelt than they did

to pelagic invertebrates when projected on the same discriminant analysis. This pattern

may be indicative of characters for these two important prey species not covered by the

proposed group labels such as ‘invertebrate’ or ‘estuarine’, and not necessarily apparent in

the PCA, thereby demonstrating the risk of generalisations based on limited samples and

reduction techniques on fatty acid data. Thus, the discriminant analyses revealed that the

signature of whelks is not wholly representative for cephalopods, and that that signatures

for euphausiids (krill) and gammarids might not be adequate for hyperiids, as these would

have failed in predicting the placement of squid and hyperiids. The result of these patterns

demonstrates the value of a very broad sampling, even if only low numbers of a variety of

species are available initially and therefore cannot be used for rigorous analysis or

predictions using discriminant functions.

54 The second reason for examining potential groupings is to propose subsequent tests

that will isolate those factors leading to the spread of scores, in particular, the intra-specific

variablity. Our results, suggest that ecological factors such as recent feeding, size, season,

and location affected the fatty acid compositions of individual specimens leading to their

separation from the expected species groupings. Both the observed group means and the

individual misclassifications provided examples of such inferences. Thus, even when

available in large sample sizes (>20) or for selected periods (i.e., autumn), pelagic and

demersal species can present wide variability, overlapping in individual scores by species,

but readily interpretable when presented as larger functional groups, such as pelagic,

demersal, fish, or invertebrate.

From the results with the repeated species, tests with more samples to control for the

size, season and location would have likely further improved class distinctions for some

species. However, from the perspective of a predator signature, further sampling may not

produce greatly different identifications beyond the significant group means already

presented here. For example, continued sampling of the pelagic species of herring, capelin,

and smelt, may be of limited value for diet-modelling beyond that which may be inferred

here when using a set of 17 fatty acids. Their fatty acid composition and the large

variability in lipid content of these species can likely be attributed to changes in their

zooplankton prey, in species, numbers, and composition, which is difficult to predict

(Henderson et al. 1984, Fraser et al. 1989, Iverson et al. 2002, Wanless et al. 2005).

Documenting patterns of similarities and differences among samples can also reveal

fine-scale trophic relationships (Iverson et al. 1997b, Iverson et al. 2002). In two separate

discriminant analyses, four turbot from the Upper Estuary and four large tomcod in spring,

specimens were classified away from their respective species and trophic groups. It is

55 possible that stomach contents of these individuals contained secondary prey items

affecting their average fatty acid signatures. Support for this hypothesis was seen when

similar-sized specimens of tomcod from the same period had stomach contents filled with

capelin, comprising a substantial fraction of the specimens’ total mass. Capelin in these

stomachs could account for the classification of larger tomcod in the pelagic rather than the

demersal species group. These misclassifications might be useful for indicating potential

trophic differences associated with a particular location or period. Hence, capelin-

influenced signatures can be anticipated during the spring spawning period, when capelin

are especially abundant and preyed upon by many larger piscivores, including marine

mammals.

Similarly-targeted tests might reveal other abundant and intermediate prey groups as

being responsible for overlapping species classifications by taking into account feeding

preferences and prey life history. Possible examples might include shrimps, krill, herring,

and sand lance rather than capelin as being consumed by halibut, plaice, and cod in summer

and autumn months. A related study of a fatty acids prey base was unable to examine

stomach contents, but also suggested that intermediate prey such as shrimps might be a

related factor in the variance and misclassification errors for piscivores such as codfishes

(Budge and Iverson 2003). In a discriminant analysis of four trophic groups in autumn,

marine shrimps overlapped considerably with flatfishes, with pelagic fishes and estuarine

sandshrimps (crangons) scoring further apart, which is suggestive of trophic similarity in

the former two groups (Annexe III e).

Having discussed the diet-related factors that could account for the observed

patterns in fatty acids and specimens, a final discussion may be warranted to show how a

study might be conducted to examine other kinds of questions. Given the differences in

56 patterns inferred by the two multivariate methods shown in this study, it may be insightful

to consider why other techniques were not considered, namely, classification trees and

clustering analysess.

Classification trees, or CART, are most useful at revealing which variables are the

most useful for segregating similar cases into known groups. In our broadly-based study,

the variables of interest were already defined as the selection of 17 diet-linked fatty acids in

relative abundance. Thus, the most important feature of CART (i.e., objective selection of

a few variables) was no longer of interest, while other useful information was not possible

when using this approach. For example, discriminant analysis, using as many variables as

permitted for given class sizes, provided statistically-robust results along with the

identification of individuals that are classified into defined categories. Moreover, we

showed that the relative position of the classified groups in discrimiant analysis can reveal

patterns that can then be subsequently tested, such as trophic groups (pelagics vs.

demersals).

Demonstrating unknown associations among cases is usually the purview of cluster

analyses. Hierarchical cluster analysis, or HCA, seeks to iteratively join cases according to

an index of similarity. Similar to PCA, HCA can reveal hidden patterns based on a number

of variables. Unlike CART, HCA is sensitive to parameters including the number of

variables, correlations among variables, data transformations, and the distance measures

used in the analysis. Furthermore, as in the case of PCA, when potential group classes such

as species overlap, the resulting associations can take many forms, with the result that it

becomes difficult to assess their significance.

For these reasons, neither CART nor HCA were used to examine overall species and

spatio-temporal variability in fatty acid signatures in this study. Nonetheless, these

57 complementary methods would be valuable in other exploratory tests. For example, CART

can be used to establish the most important fatty acids that separate intraspecific groups by

year, season, and location—even when only small sample sizes are available (i.e., <6), the

cases are similar (within-species differences), and the important variables are unknown.

This type of analysis would be more appropriate for a detailed study restricted towards

understanding the biology and diet of a species, such as capelin, because such small-scale

differences would not be considered in predator diet-modeling. Similarly, cluster analysis

can reveal associations if samples were to differ on a smaller group of variables of interest,

as when using PCA with a larger number of fatty acids to produce a reduced set of

independent factors. An example here might be to examine in greater detail the effect of

size-classes of a species on compositions, as between sardines and herring.

2.4. Conclusion

We have documented the lipid contents and fatty acid compositions of a wide range

of species, with an additional sampling effort aimed at assessing spatial and temporal

variability of potential prey of beluga. By highlighting the similarities and differences

between specimens when using a selection of fatty acids, both of the multivariate methods

we used revealed their capacities to distinguish among the signatures of prey species.

Perhaps of greater interest from this exploratory study (i.e., when restricted in the number

of variables and samples) was the suggestion of apparent functional groups among species

that were both expected, as in the case of herring and capelin as pelagic fishes in all

58 analyses, but also novel, such as the intermediate scores of smelt between pelagic and

demersal species in discriminant analysis but not in PCA.

The discrimination of prey species achieved using the diet-linked fatty acids

suggests these signatures have the potential to trace trophic linkages. However, of some

concern was the similarity (relative proximity of scores) among specimens of pelagic

species such as herring and capelin, and the wider variability (dispersion of scores) seen

among demersal species like tomcod and flatfishes, but also among smelt and invertebrates.

The detailed examination of fatty acid compositions reiterated the need for additional

considerations such as an awareness of prey species distributions, weighting by lipid

contents, and the differences in output of multivariate analyses when evaluating fatty acid

signatures for trophic-level studies.

Overall, our analyses improved the understanding of the fatty acid compositions of

prey items available in the local food web for the St. Lawrence beluga whale. If the use of

sampled beluga carcasses can be validated (to compensate for quality issues and calibrate

for predator biology), the information available in this prey base will lead to better

knowledge of diet linkages through the use of models such as QFASA, thereby helping to

shed new light on persistent questions, including the source of environmental contaminants

in predators like belugas. Finally, information derived from the prey base and future diet

models would confirm the importance of presumed prey species of belugas, and therefore

help identify the habitats upon which they depend, that are thus deserving of protection to

ensure the viability of this unique population of whales.

59 Table 2.1. Summary of samples collected in 2000-2002 by species, season and sub-

region: Upper Estuary (A), Lower Estuary (B), Sept-Iles (C), and northern

Gulf (D). Seasons are spring (April–June), summer (July–Aug.), autumn

(Sept.–Oct.) and winter (Feb.). Samples were analysed either as whole

specimens or as a pooled set of specimens in the case of zooplankton, i.e,

amphipods, copepods, and euphausiids (krill).

n by sub-region Species name Common name Season Total A B C D Alosa sapidissima American shad summer 13 13 - - -Amblyraja radiata Thorny skate summer 6 5 1 - - autumn 6 6 - - - Ammodytes sp. Sand lance spring 18 - - 6 - summer 12 12 6 6 - winter 12 - 12 - -Anarhichas lupus Atlantic wolffish summer 8 - - - 8Anarhichas minor Striped wolffish summer 1 - - - 1Anguilla rostrata American eel autumn 14 14 - - -Arctozenus risso White barracudina spring 6 - 6 - - summer 12 - 12 - -Argis dentata Argid shrimp autumn 12 - 12 - -Bathypolypus arcticus Boreal octopus spring 5 - - 5 -Boreogadus saida Arctic cod summer 4 - 4 - -Buccinum undatum Waved whelk spring 18 6 6 6 - summer 6 - - - 6Calanus hyperboreus Copepod spring 1 - 1 - - summer 11 - 11 - -Chionoecetes opilio Snow crab spring 6 - 6 - -Colus sp. Whelk autumn 2 - 2 - - Clupea harengus Atlantic herring spring 48 24 24 - - summer 42 30 - 12 - autumn 36 36 - - -Coregonus sp. Whitefish spring 2 2 - - - autumn 3 3 - - -Crangon septemspinosa Sand shrimp spring 6 6 - - - summer 6 6 - - - autumn 17 17 - - -Cyclopterus lumpus Lumpfish spring 7 - 7 - - summer 5 - 2 3 -

60 Table 2.1 (cont’d)

n by sub-region Species name Common name Season Total A B C D Enchelyopus cimbrius Fourbeard rockling summer 3 - 3 - - autumn 4 - 4 - -Eualus macilentus Eualid shrimp spring 9 - 9 - - summer 3 - 3 - - autumn 6 - 6 - - Gadus morhua Atlantic cod spring 5 - 5 - - summer 9 - 9 - - Gammarus sp. Gammarid amphipod spring 4 - - 4 - Glyptocephalus cynoglossus Witch flounder summer 5 5 - - - Gymnacanthus tricuspis Arctic staghorn sculpin summer 7 - 1 - 6 Hippoglossoides platessoides American plaice spring 6 - 4 2 - summer 22 5 9 8 - autumn 12 - 10 2 - Hyas araneus Toad crab summer 1 - 1 - - Illex illecebrosus Shortfin squid summer 6 - - - 6 Leptagonus decagonus Atlantic poacher summer 8 - 8 - - autumn 4 - 4 - - Limanda ferruginea Yellowtail flounder summer 1 - 1 - - Liparis gibbus Snailfish spring 6 6 - - - summer 7 - 7 - - autumn 5 - 5 - - Lophius americanus Monkfish summer 4 - - - 4 Leptoclinus maculatus Daubed shanny summer 6 - 6 - - Lycodes vahlii Checker eelpout summer 1 - 1 - - autumn 7 - 7 - - Mallotus villosus Capelin spring 71 23 18 30 - summer 17 11 6 - - autumn 7 7 - - - Meganyctiphanes norvegica Northern krill spring 27 - 21 6 - summer 6 - 6 - - Melanogrammus aeglefinus Haddock summer 2 - 2 - - Merluccius biinearis Silver hake summer 6 - 6 - - Microgadus tomcod Atlantic tomcod spring 18 18 - - - summer 12 12 - - - autumn 9 9 - - - Myoxocephalus scorpius Shorthorn sculpin spring 1 - 1 - - summer 7 - 7 - - Myxine glutinosa Hagfish summer 12 - 12 - -

61 Table 2.1 (cont’d)

n by sub-region Species name Common name Season Total A B C D Nereis virens Polychaete worm spring 6 6 - - - autumn 6 6 - - - Nezumia bairdii Marlin-spike summer 6 6 - - - Osmerus mordax Rainbow smelt spring 42 36 6 - - summer 28 12 16 - - autumn 57 45 12 - - Pandalus borealis Northern shrimp summer 6 - 6 - - autumn 6 - 6 - - Pandalus montagui Striped shrimp spring 9 - 9 - - summer 3 - 3 - - autumn 6 - 6 - - Petromyzontidae Lampreys autumn 3 2 1 - - Pseudopleuronectes Winter flounder spring 12 12 - - - americanus summer 18 16 2 - - autumn 13 12 1 - - Pollachius virens Pollock summer 6 3 1 - 2 Reinhardtius hippoglossoides Greenland halibut spring 6 - 6 - - (turbot) summer 19 4 15 - - autumn 1 - 1 - - Salmo salar Atlantic salmon spring 15 - 15 - - Sclerocrangon boreas Scampi shrimp summer 6 - - - 6 Sebastes sp. Deepwater redfish spring 12 - 12 - - Semirossia tenera Bobtail squid summer 7 - - - 7 Themisto libellula Hyperiid amphipod spring 8 - 3 5 - summer 3 - 3 - - autumn 5 - 5 - - Thysanoessa raschii Krill autumn 5 - 5 - - Triglops murrayi Moustache sculpin summer 9 - 9 - - Urophycis chesteri Longfin hake summer 6 - 6 - - Urophycis tenuis White hake summer 7 - 7 - - Zoarces americanus Ocean pout autumn 1 1 - - -

62 Table 2.2. (FOLLOWING PAGES)

Composition (% of total by weight) for 25 fatty acids, each averaging >0.4%,

among all species groups analysed ( = fatty acid not retained in statistical

analyses). Shad, herring, and salmon were each segregated into small (<18

cm) and large (>18 cm) groups. Specimen standard length, wet weight, and

lipid content are also presented. Values are means ± SEM.

Table 2.2.

Common name Shad-sardines Shad (large) Sand lance Atlantic wolffish Spotted wolffish Species group Alosa sapidissima Alosa sapidissima-L Ammodytes sp. Anarhichas lupus Anarhichas minor

n 12 1 42 8 1 c14:0 2.57 ± 0.38 4.93 3.97 ± 0.19 2.72 ± 0.35 1.87 c16:0 20.54 ± 0.26 14.21 14.66 ± 0.59 13.51 ± 0.75 11.04

c16:1n-7 4.89 ± 0.92 3.05 6.77 ± 0.31 8.89 ± 0.98 5.64 c16:3n-6 0.65 ± 0.18 0.33 0.57 ± 0.04 0.15 ± 0.02 0.23 c16:4n-1 0.22 ± 0.01 0.23 0.85 ± 0.08 0.19 ± 0.04 0.24

c17:0 0.74 ± 0.02 0.40 1.17 ± 0.18 0.36 ± 0.06 0.50 c18:0 6.10 ± 0.21 2.52 2.74 ± 0.18 3.43 ± 0.31 4.79

c18:1n-11 0.05 ± 0.01 0.89 0.39 ± 0.03 0.66 ± 0.07 0.67 c18:1n-9 8.03 ± 0.32 9.08 6.91 ± 0.37 17.18 ± 2.25 10.86 c18:1n-7 3.94 ± 0.07 3.13 3.61 ± 0.23 6.37 ± 0.42 9.23 c18:1n-5 0.11 ± 0.01 0.45 0.52 ± 0.02 0.66 ± 0.05 0.98 c18:2n-6 2.02 ± 0.24 0.99 0.86 ± 0.04 0.66 ± 0.02 0.59 c18:3n-3 1.07 ± 0.19 0.75 0.34 ± 0.02 0.20 ± 0.02 0.16 c18:4n-3 0.69 ± 0.12 1.52 1.03 ± 0.07 0.33 ± 0.06 0.45

c20:1n-11 0.12 ± 0.02 1.62 0.44 ± 0.03 2.54 ± 0.84 2.65 c20:1n-9 0.36 ± 0.06 15.58 8.37 ± 0.76 2.99 ± 0.37 3.38 c20:1n-7 0.22 ± 0.01 0.26 0.61 ± 0.04 0.96 ± 0.17 1.39 c20:2n-6 0.19 ± 0.02 0.24 0.15 ± 0.01 0.31 ± 0.04 0.36 c20:4n-6 3.57 ± 0.26 0.60 0.57 ± 0.06 4.28 ± 1.29 8.96 c20:5n-3 8.45 ± 0.23 6.08 12.71 ± 0.33 11.21 ± 0.84 12.49

c22:1n-11 0.20 ± 0.14 13.50 9.46 ± 0.88 1.87 ± 0.33 1.24 c22:1n-9 0.05 ± 0.01 0.87 1.13 ± 0.10 0.43 ± 0.04 0.38 c22:5n-3 4.08 ± 0.12 2.08 0.88 ± 0.04 1.11 ± 0.15 1.04 c22:6n-3 22.29 ± 1.72 10.31 14.65 ± 0.92 10.13 ± 0.42 10.05 c24:1n-9 0.48 ± 0.04 0.79 0.94 ± 0.04 0.98 ± 0.10 0.83

∑ 91.63 94.36 94.28 92.10 89.95 length (cm) 10.26 ± 0.12 46.40 11.48 ± 0.38 34.61 ± 3.23 38.60 weight (g) 8.90 ± 0.31 1543.90 4.04 ± 0.40 619.26 ± 147.07 698.50 lipid (%) 1.47 ± 0.14 7.55 3.37 ± 0.28 1.90 ± 0.29 1.26

64

Table 2.2. (cont’d) Common name Thorny skate American eel White barracudina Argid shrimp Boreal octopus Species group Amblyraja radiata Anguilla rostrata Arctozenus risso Argis dentata Bathypolypus arcticus

n 12 14 18 12 5 c14:0 0.96 ± 0.14 3.56 ± 0.14 5.74 ± 0.32 1.47 ± 0.09 1.22 ± 0.06 c16:0 16.08 ± 0.56 17.75 ± 0.38 16.03 ± 0.45 13.34 ± 0.16 15.97 ± 0.97

c16:1n-7 4.63 ± 0.36 10.36 ± 0.21 9.76 ± 0.18 9.07 ± 0.47 1.48 ± 0.66 c16:3n-6 0.12 ± 0.02 0.26 ± 0.03 0.60 ± 0.02 0.32 ± 0.02 0.04 ± 0.01 c16:4n-1 0.17 ± 0.03 0.12 ± 0.01 0.30 ± 0.01 0.23 ± 0.03 0.17 ± 0.02

c17:0 0.58 ± 0.08 0.51 ± 0.02 0.30 ± 0.08 0.72 ± 0.04 0.85 ± 0.03 c18:0 4.15 ± 0.40 2.85 ± 0.08 2.42 ± 0.09 2.44 ± 0.07 3.93 ± 0.17

c18:1n-11 0.94 ± 0.15 0.02 ± 0.01 0.15 ± 0.01 0.14 ± 0.01 0.34 ± 0.06 c18:1n-9 12.24 ± 0.30 28.78 ± 0.79 18.68 ± 0.75 10.13 ± 0.24 3.32 ± 0.74 c18:1n-7 7.09 ± 0.24 5.25 ± 0.31 4.82 ± 0.15 9.80 ± 0.14 3.85 ± 0.57 c18:1n-5 0.53 ± 0.04 0.23 ± 0.02 0.42 ± 0.02 0.77 ± 0.02 0.64 ± 0.08 c18:2n-6 1.31 ± 0.07 3.25 ± 0.32 0.65 ± 0.02 0.73 ± 0.02 0.57 ± 0.07 c18:3n-3 0.24 ± 0.02 2.25 ± 0.23 0.32 ± 0.01 0.16 ± 0.01 0.11 ± 0.03 c18:4n-3 0.18 ± 0.04 0.18 ± 0.06 0.70 ± 0.04 0.45 ± 0.06 0.08 ± 0.03 c20:1n-11 0.94 ± 0.08 0.51 ± 0.07 0.47 ± 0.03 1.24 ± 0.13 0.99 ± 0.22 c20:1n-9 4.42 ± 1.13 1.10 ± 0.05 8.00 ± 0.50 0.72 ± 0.03 6.85 ± 0.59 c20:1n-7 1.05 ± 0.10 0.15 ± 0.01 0.55 ± 0.02 1.70 ± 0.09 1.36 ± 0.33 c20:2n-6 0.54 ± 0.05 0.60 ± 0.04 0.14 ± 0.00 0.42 ± 0.01 0.74 ± 0.09 c20:4n-6 4.27 ± 0.53 3.20 ± 0.18 0.25 ± 0.02 1.87 ± 0.06 4.30 ± 0.31 c20:5n-3 8.08 ± 0.55 2.77 ± 0.17 7.54 ± 0.31 20.86 ± 0.45 19.86 ± 0.74 c22:1n-11 3.20 ± 1.13 0.00 ± 0.00 9.82 ± 0.70 0.14 ± 0.02 1.43 ± 0.41 c22:1n-9 0.64 ± 0.10 0.05 ± 0.00 1.31 ± 0.07 0.22 ± 0.02 1.62 ± 0.24 c22:5n-3 2.48 ± 0.31 2.52 ± 0.14 0.72 ± 0.04 2.29 ± 0.12 1.22 ± 0.18 c22:6n-3 18.86 ± 0.81 3.69 ± 0.37 5.05 ± 0.22 10.49 ± 0.32 21.25 ± 1.13 c24:1n-9 0.36 ± 0.03 0.07 ± <0.01 0.55 ± 0.03 0.13 ± 0.01 0.32 ± 0.05

∑ 94.05 90.03 94.78 89.72 92.23 length (cm) 21.07 ± 1.09 79.49 ± 1.58 24.11 ± 0.35 8.60 ± 0.22 15.38 ± 3.23 weight (g) 76.31 ± 15.03 1094.90 ± 60.52 24.61 ± 1.65 6.01 ± 0.50 51.68 ± 24.79 lipid (%) 1.57 ± 0.20 13.29 ± 0.34 12.53 ± 0.70 1.99 ± 0.13 1.13 ± 0.12

65

Table 2.2. (cont’d) Common name Arctic cod Waved whelk Copepod Snow crab Herring-sardines Species group Boreogadus saida Buccinum undatum Calanus hyperboreus Chionoecetes opilio Clupea harengus

n 4 24 12 6 42 c14:0 2.83 ± 0.13 2.08 ± 0.12 3.37 ± 0.54 1.33 ± 0.14 5.17 ± 0.15 c16:0 18.89 ± 0.32 10.97 ± 0.26 4.87 ± 0.21 11.39 ± 0.09 13.74 ± 0.44

c16:1n-7 8.22 ± 0.77 2.51 ± 0.25 8.82 ± 0.11 5.32 ± 0.44 6.64 ± 0.24 c16:3n-6 0.87 ± 0.12 0.12 ± 0.02 0.93 ± 0.04 0.17 ± 0.02 0.73 ± 0.03 c16:4n-1 0.30 ± 0.03 0.19 ± 0.02 1.96 ± 0.16 0.29 ± 0.03 0.83 ± 0.05

c17:0 0.10 ± 0.00 0.60 ± 0.03 0.04 ± 0.00 0.45 ± 0.05 0.49 ± 0.08 c18:0 4.10 ± 0.31 5.34 ± 0.29 0.32 ± 0.02 2.49 ± 0.15 1.53 ± 0.08

c18:1n-11 0.60 ± 0.16 0.48 ± 0.10 0.06 ± 0.03 0.53 ± 0.05 0.41 ± 0.03 c18:1n-9 9.59 ± 0.36 4.64 ± 0.28 1.94 ± 0.36 10.79 ± 0.75 6.74 ± 0.51 c18:1n-7 7.79 ± 0.62 4.74 ± 0.29 1.37 ± 0.04 6.80 ± 0.20 3.18 ± 0.19 c18:1n-5 0.27 ± 0.02 0.43 ± 0.05 0.52 ± 0.01 0.73 ± 0.05 0.45 ± 0.02 c18:2n-6 0.57 ± 0.02 1.38 ± 0.07 0.23 ± 0.02 0.76 ± 0.04 0.56 ± 0.02 c18:3n-3 0.24 ± 0.01 0.45 ± 0.06 0.10 ± 0.01 0.20 ± 0.01 0.25 ± 0.01 c18:4n-3 0.83 ± 0.05 0.44 ± 0.06 1.49 ± 0.12 0.28 ± 0.05 0.85 ± 0.07

c20:1n-11 0.43 ± 0.08 5.67 ± 0.23 0.57 ± 0.11 1.71 ± 0.08 0.70 ± 0.03 c20:1n-9 3.03 ± 0.46 3.97 ± 0.41 21.65 ± 0.52 4.92 ± 0.43 12.08 ± 0.72 c20:1n-7 0.41 ± 0.02 3.46 ± 0.28 2.56 ± 0.15 1.68 ± 0.19 0.58 ± 0.03 c20:2n-6 0.23 ± 0.00 3.86 ± 0.30 0.07 ± 0.00 0.63 ± 0.05 0.13 ± 0.01 c20:4n-6 0.60 ± 0.05 4.60 ± 0.63 0.22 ± 0.01 3.27 ± 0.25 0.41 ± 0.03 c20:5n-3 17.19 ± 0.32 20.91 ± 0.68 7.29 ± 0.28 19.18 ± 1.11 9.05 ± 0.41

c22:1n-11 3.06 ± 0.54 2.57 ± 0.64 29.14 ± 0.41 5.26 ± 0.73 15.41 ± 0.77 c22:1n-9 0.50 ± 0.06 0.41 ± 0.06 3.43 ± 0.20 0.65 ± 0.06 1.15 ± 0.07 c22:5n-3 0.69 ± 0.06 5.09 ± 0.37 0.56 ± 0.03 1.11 ± 0.07 0.81 ± 0.03 c22:6n-3 14.22 ± 1.17 7.60 ± 0.65 2.76 ± 0.17 12.83 ± 0.63 12.32 ± 0.74 c24:1n-9 0.61 ± 0.12 0.08 ± 0.02 0.41 ± 0.05 0.31 ± 0.02 0.73 ± 0.03

∑ 95.66 92.50 94.31 92.79 94.26 length (cm) 10.63 ± 0.17 7.87 ± 0.14 - - - 6.82 ± 0.24 13.46 ± 0.26 weight (g) 8.77 ± 0.39 61.42 ± 3.73 - - - 104.90 ± 8.05 20.59 ± 0.98 lipid (%) 2.84 ± 0.37 1.30 ± 0.12 14.81 ± 1.49 1.06 ± 0.16 3.94 ± 0.30

66

Table 2.2. (cont’d) Common name Herring (large) Whelk Whitefish Sand shrimp Lumpfish Species group Clupea harengus-L Colus sp. Coregonus sp. Crangon septemspinosa Cyclopterus lumpus

n 84 2 5 29 12 c14:0 4.58 ± 0.10 1.65 ± 0.06 1.64 ± 0.26 1.37 ± 0.10 3.23 ± 0.36 c16:0 11.41 ± 0.24 8.59 ± 0.32 16.80 ± 0.68 15.39 ± 0.32 11.33 ± 0.31

c16:1n-7 6.53 ± 0.19 1.64 ± 0.01 8.76 ± 1.88 5.01 ± 0.37 4.25 ± 0.63 c16:3n-6 0.48 ± 0.02 0.06 ± 0.00 0.34 ± 0.08 0.18 ± 0.04 0.24 ± 0.03 c16:4n-1 0.62 ± 0.03 0.07 ± 0.01 0.25 ± 0.12 0.19 ± 0.03 0.11 ± 0.02

c17:0 0.26 ± 0.04 0.71 ± 0.08 0.53 ± 0.09 0.95 ± 0.08 0.29 ± 0.02 c18:0 1.17 ± 0.04 5.24 ± 0.40 3.83 ± 0.30 3.40 ± 0.20 3.00 ± 0.22

c18:1n-11 0.69 ± 0.03 1.19 ± 0.75 0.76 ± 0.35 0.32 ± 0.05 3.69 ± 0.32 c18:1n-9 7.45 ± 0.33 5.96 ± 0.51 17.17 ± 1.77 10.30 ± 0.28 13.19 ± 0.49 c18:1n-7 2.75 ± 0.10 5.26 ± 0.25 5.06 ± 0.56 4.98 ± 0.16 4.29 ± 0.12 c18:1n-5 0.45 ± 0.01 0.81 ± 0.11 0.38 ± 0.06 0.47 ± 0.04 0.60 ± 0.03 c18:2n-6 0.58 ± 0.02 1.61 ± 0.24 2.16 ± 0.41 1.47 ± 0.08 1.02 ± 0.04 c18:3n-3 0.23 ± 0.01 0.15 ± 0.02 1.47 ± 0.35 0.51 ± 0.05 0.41 ± 0.03 c18:4n-3 0.66 ± 0.04 0.07 ± 0.02 0.39 ± 0.11 0.19 ± 0.04 0.71 ± 0.14

c20:1n-11 1.28 ± 0.08 7.91 ± 0.02 1.02 ± 0.38 1.33 ± 0.08 1.25 ± 0.17 c20:1n-9 15.03 ± 0.35 4.36 ± 1.70 1.63 ± 0.68 3.07 ± 0.62 7.28 ± 0.72 c20:1n-7 0.99 ± 0.03 3.40 ± 0.62 0.65 ± 0.13 1.45 ± 0.10 1.07 ± 0.10 c20:2n-6 0.10 ± 0.00 3.32 ± 0.25 0.77 ± 0.13 0.55 ± 0.03 0.25 ± 0.01 c20:4n-6 0.30 ± 0.01 6.41 ± 1.00 3.53 ± 1.05 3.32 ± 0.25 0.70 ± 0.11 c20:5n-3 6.01 ± 0.16 15.04 ± 5.63 7.76 ± 0.85 19.81 ± 0.63 13.45 ± 0.48

c22:1n-11 22.80 ± 0.47 2.01 ± 1.25 0.88 ± 0.65 3.06 ± 0.75 3.67 ± 0.47 c22:1n-9 2.06 ± 0.06 0.39 ± 0.15 0.23 ± 0.08 0.47 ± 0.10 1.70 ± 0.25 c22:5n-3 0.68 ± 0.02 6.31 ± 0.73 2.53 ± 0.22 1.88 ± 0.13 1.56 ± 0.16 c22:6n-3 8.08 ± 0.30 8.53 ± 0.44 11.92 ± 3.84 11.92 ± 0.54 16.63 ± 1.40 c24:1n-9 0.64 ± 0.02 0.22 ± 0.05 0.50 ± 0.23 0.33 ± 0.03 0.51 ± 0.06

∑ 95.21 90.73 90.67 91.62 94.00 length (cm) 24.83 ± 0.27 10.80 ± 2.00 37.84 ± 2.58 5.76 ± 0.13 26.65 ± 1.66 weight (g) 159.92 ± 5.11 65.55 ± 17.88 612.67 ± 97.93 1.68 ± 0.14 887.33 ± 106.27 lipid (%) 7.35 ± 0.40 0.81 ± 0.03 3.93 ± 1.22 1.42 ± 0.10 2.41 ± 0.52

67

Table 2.2. (cont’d) Common name Fourbeard rockling Eualid shrimp Atlantic cod Gammarid amphipod Witch flounder Species group Enchelyopus cimbrius Eualus macilentus Gadus morhua Gammarus sp. Glyptocephalus cynoglossus

n 7 18 14 4 5 c14:0 1.28 ± 0.13 1.71 ± 0.16 1.99 ± 0.19 2.14 ± 0.28 3.40 ± 0.21 c16:0 16.47 ± 0.59 15.60 ± 0.34 15.07 ± 0.37 13.68 ± 0.79 15.48 ± 0.23

c16:1n-7 3.66 ± 0.55 7.53 ± 0.80 4.79 ± 0.37 10.20 ± 1.27 8.88 ± 1.03 c16:3n-6 0.14 ± 0.02 0.42 ± 0.08 0.25 ± 0.03 0.21 ± 0.03 0.26 ± 0.01 c16:4n-1 0.39 ± 0.12 0.31 ± 0.08 0.08 ± 0.02 0.05 ± 0.01 0.32 ± 0.09

c17:0 0.50 ± 0.08 0.34 ± 0.03 0.18 ± 0.01 0.19 ± 0.06 0.70 ± 0.09 c18:0 4.38 ± 0.28 2.10 ± 0.10 3.17 ± 0.10 0.84 ± 0.19 3.48 ± 0.17

c18:1n-11 1.21 ± 0.12 0.39 ± 0.11 1.83 ± 0.14 0.53 ± 0.06 0.62 ± 0.12 c18:1n-9 11.83 ± 0.89 10.95 ± 0.48 10.37 ± 0.47 22.98 ± 3.03 6.68 ± 0.91 c18:1n-7 5.11 ± 0.32 10.00 ± 0.37 4.91 ± 0.25 5.76 ± 0.68 5.32 ± 0.45 c18:1n-5 0.43 ± 0.03 0.62 ± 0.02 0.37 ± 0.02 0.69 ± 0.14 0.55 ± 0.08 c18:2n-6 0.56 ± 0.04 0.91 ± 0.05 0.56 ± 0.02 1.29 ± 0.10 1.00 ± 0.03 c18:3n-3 0.10 ± 0.01 0.28 ± 0.01 0.17 ± 0.02 0.38 ± 0.06 0.23 ± 0.03 c18:4n-3 0.20 ± 0.03 0.48 ± 0.09 0.29 ± 0.04 0.51 ± 0.15 0.24 ± 0.05

c20:1n-11 0.76 ± 0.06 0.54 ± 0.05 1.15 ± 0.08 1.84 ± 0.13 2.11 ± 0.14 c20:1n-9 2.90 ± 0.23 2.15 ± 0.48 5.53 ± 0.48 4.84 ± 0.24 3.52 ± 0.59 c20:1n-7 0.91 ± 0.10 1.06 ± 0.09 0.51 ± 0.03 1.36 ± 0.36 2.38 ± 0.24 c20:2n-6 0.25 ± 0.01 0.36 ± 0.01 0.21 ± 0.01 0.47 ± 0.12 0.57 ± 0.06 c20:4n-6 3.17 ± 0.23 3.59 ± 0.44 1.92 ± 0.17 1.59 ± 0.32 2.85 ± 0.31 c20:5n-3 11.55 ± 0.38 17.69 ± 0.76 13.18 ± 0.47 11.83 ± 0.47 9.63 ± 0.28

c22:1n-11 0.93 ± 0.17 2.43 ± 0.80 3.98 ± 0.47 2.23 ± 0.64 3.22 ± 0.95 c22:1n-9 0.36 ± 0.03 0.38 ± 0.06 0.55 ± 0.04 0.42 ± 0.08 0.64 ± 0.14 c22:5n-3 1.58 ± 0.04 0.88 ± 0.05 1.28 ± 0.05 1.14 ± 0.28 2.86 ± 0.26 c22:6n-3 22.28 ± 1.49 11.42 ± 0.77 22.46 ± 1.00 9.47 ± 0.38 9.68 ± 1.08 c24:1n-9 2.23 ± 0.37 0.33 ± 0.02 0.92 ± 0.05 0.15 ± 0.05 0.91 ± 0.24

∑ 91.30 92.16 94.87 94.66 84.85 length (cm) 20.31 ± 0.75 - - - 32.79 ± 1.59 - - - 25.02 ± 0.27 weight (g) 44.25 ± 10.01 1.21 ± 0.14 341.11 ± 46.03 0.74 ± 0.11 166.53 ± 8.07 lipid (%) 1.02 ± 0.12 2.18 ± 0.22 1.48 ± 0.11 3.20 ± 0.77 1.43 ± 0.24

68

Table 2.2. (cont’d) Common name Arctic staghorn sculpin American plaice Toad crab Shortfin squid Atlantic poacher Species group Gymnacanthus tricuspis Hippoglossoides platessoides Hyas araneus Illex illecebrosus Leptagonus decagonus

n 7 40 1 6 12 c14:0 2.41 ± 0.12 3.13 ± 0.19 0.65 3.14 ± 0.20 2.55 ± 0.15 c16:0 14.55 ± 0.43 14.58 ± 0.26 7.87 15.52 ± 0.41 13.10 ± 0.32

c16:1n-7 7.46 ± 1.00 7.22 ± 0.43 2.38 3.47 ± 0.46 10.27 ± 0.46 c16:3n-6 0.36 ± 0.07 0.46 ± 0.04 0.04 0.16 ± 0.02 0.42 ± 0.02 c16:4n-1 0.55 ± 0.06 0.74 ± 0.06 0.19 0.10 ± 0.03 0.23 ± 0.03

c17:0 0.39 ± 0.02 0.68 ± 0.07 0.83 0.45 ± 0.03 0.49 ± 0.05 c18:0 3.67 ± 0.26 3.25 ± 0.16 4.34 2.52 ± 0.11 2.07 ± 0.18

c18:1n-11 0.74 ± 0.13 0.56 ± 0.06 0.16 0.64 ± 0.08 1.18 ± 0.10 c18:1n-9 10.77 ± 1.38 9.26 ± 0.31 8.73 7.88 ± 0.96 13.52 ± 0.57 c18:1n-7 8.25 ± 0.27 5.49 ± 0.18 10.16 3.13 ± 0.15 8.31 ± 0.16 c18:1n-5 0.57 ± 0.04 0.38 ± 0.01 0.71 0.63 ± 0.04 0.86 ± 0.04 c18:2n-6 0.73 ± 0.03 0.77 ± 0.02 0.99 0.85 ± 0.14 1.29 ± 0.06 c18:3n-3 0.26 ± 0.02 0.17 ± 0.01 0.09 0.45 ± 0.04 0.30 ± 0.01 c18:4n-3 0.81 ± 0.09 0.59 ± 0.06 0.08 0.73 ± 0.10 0.45 ± 0.03

c20:1n-11 0.66 ± 0.04 0.86 ± 0.09 1.92 1.45 ± 0.11 0.76 ± 0.05 c20:1n-9 2.00 ± 0.24 4.58 ± 0.52 1.43 9.41 ± 0.71 4.35 ± 0.38 c20:1n-7 1.06 ± 0.08 1.55 ± 0.11 1.90 0.53 ± 0.03 1.16 ± 0.07 c20:2n-6 0.29 ± 0.03 0.25 ± 0.01 1.97 0.36 ± 0.01 0.28 ± 0.01 c20:4n-6 1.86 ± 0.16 2.37 ± 0.19 11.34 0.56 ± 0.02 1.46 ± 0.13 c20:5n-3 17.17 ± 0.65 13.53 ± 0.51 22.09 12.40 ± 0.25 12.97 ± 0.25

c22:1n-11 0.65 ± 0.13 3.48 ± 0.49 0.39 6.49 ± 0.54 2.59 ± 0.26 c22:1n-9 0.36 ± 0.04 0.59 ± 0.07 0.26 0.83 ± 0.06 0.63 ± 0.04 c22:5n-3 2.04 ± 0.19 2.23 ± 0.07 1.89 0.78 ± 0.10 1.20 ± 0.04 c22:6n-3 13.52 ± 1.15 15.03 ± 0.64 8.38 21.93 ± 1.05 12.27 ± 0.72 c24:1n-9 0.67 ± 0.11 0.87 ± 0.04 0.22 0.50 ± 0.02 0.64 ± 0.04

∑ 91.25 91.80 89.23 94.44 92.74 length (cm) 13.40 ± 1.05 24.64 ± 0.90 13.10 154.40 ± 16.71 17.98 ± 0.60 weight (g) 39.35 ± 8.78 170.91 ± 19.64 172.39 97.87 ± 4.08 20.53 ± 2.23 lipid (%) 1.94 ± 0.30 1.54 ± 0.11 0.59 3.73 ± 0.23 3.85 ± 0.50

69

Table 2.2. (cont’d)

Common name Daubed shanny Yellowtail flounder Snailfish Monkfish Checker eelpout Species group Leptoclinus maculatus Limanda ferruginea Liparis gibbus Lophius americanus Lycodes vahlii

n 6 1 18 4 8 c14:0 3.05 ± 0.11 2.42 2.51 ± 0.26 3.47 ± 0.37 2.44 ± 0.28 c16:0 16.02 ± 0.40 17.56 16.30 ± 0.26 13.88 ± 0.58 13.88 ± 0.44

c16:1n-7 13.05 ± 0.49 14.54 7.99 ± 0.95 13.05 ± 1.04 7.46 ± 0.94 c16:3n-6 0.51 ± 0.04 0.20 0.66 ± 0.11 0.34 ± 0.03 0.40 ± 0.06 c16:4n-1 0.36 ± 0.06 0.14 0.41 ± 0.02 0.21 ± 0.04 0.26 ± 0.03

c17:0 0.37 ± 0.06 0.20 1.18 ± 0.13 0.17 ± 0.01 0.44 ± 0.03 c18:0 3.18 ± 0.22 2.61 3.78 ± 0.23 2.59 ± 0.27 3.70 ± 0.48

c18:1n-11 0.24 ± 0.03 0.31 0.52 ± 0.09 1.29 ± 0.23 1.00 ± 0.10 c18:1n-9 15.43 ± 0.58 13.29 15.61 ± 0.43 13.24 ± 1.10 10.78 ± 0.22 c18:1n-7 6.16 ± 0.23 4.31 7.73 ± 0.58 4.92 ± 0.09 5.97 ± 0.18 c18:1n-5 0.55 ± 0.04 0.35 0.33 ± 0.02 0.55 ± 0.02 0.66 ± 0.04 c18:2n-6 0.59 ± 0.05 0.84 1.32 ± 0.21 0.97 ± 0.04 1.23 ± 0.07 c18:3n-3 0.23 ± 0.03 0.55 0.53 ± 0.08 0.28 ± 0.02 0.31 ± 0.04 c18:4n-3 0.57 ± 0.04 0.36 0.84 ± 0.10 0.65 ± 0.06 0.49 ± 0.07 c20:1n-11 0.79 ± 0.08 1.02 0.34 ± 0.06 0.95 ± 0.10 1.01 ± 0.09 c20:1n-9 2.10 ± 0.13 1.57 1.78 ± 0.21 8.93 ± 0.97 4.71 ± 0.51 c20:1n-7 2.47 ± 0.44 0.97 0.39 ± 0.03 0.91 ± 0.08 1.53 ± 0.19 c20:2n-6 0.27 ± 0.02 0.67 0.38 ± 0.06 0.26 ± 0.00 0.42 ± 0.03 c20:4n-6 1.04 ± 0.05 2.06 2.84 ± 0.78 1.44 ± 0.34 3.43 ± 0.53 c20:5n-3 12.21 ± 0.30 13.33 17.12 ± 0.32 5.93 ± 0.77 13.73 ± 0.39 c22:1n-11 2.21 ± 0.21 0.60 0.39 ± 0.08 5.79 ± 0.82 2.25 ± 0.39 c22:1n-9 0.37 ± 0.02 0.26 0.13 ± 0.01 1.37 ± 0.15 0.47 ± 0.06 c22:5n-3 1.50 ± 0.09 4.59 1.48 ± 0.38 1.28 ± 0.15 1.03 ± 0.07 c22:6n-3 8.00 ± 0.47 8.73 9.41 ± 0.25 9.68 ± 1.70 13.26 ± 1.04 c24:1n-9 0.39 ± 0.04 0.41 0.53 ± 0.10 0.95 ± 0.09 1.10 ± 0.21

∑ 91.30 91.43 94.49 93.06 91.97 length (cm) 11.90 ± 0.06 33.80 12.04 ± 0.53 40.95 ± 2.55 25.70 ± 0.65 weight (g) 6.65 ± 0.36 529.68 30.87 ± 3.12 1789.45 ± 242.24 62.37 ± 4.68 lipid (%) 4.90 ± 0.31 2.42 2.38 ± 0.23 3.56 ± 0.90 1.59 ± 0.29

70

Table 2.2. (cont’d) Common name Capelin Northern krill Haddock Silver hake Tomcod

Species-size Mallotus villosus Meganyctiphanes norvegica Melanogrammus aeglefinus Merluccius biinearis Microgadus tomcod n 95 33 2 6 39

c14:0 4.15 ± 0.17 4.27 ± 0.11 3.20 ± 0.28 3.11 ± 0.12 1.98 ± 0.12 c16:0 15.53 ± 0.34 12.95 ± 0.27 13.22 ± 0.48 16.58 ± 0.45 15.27 ± 0.21

c16:1n-7 6.56 ± 0.33 8.67 ± 0.14 5.81 ± 0.06 5.31 ± 0.26 5.20 ± 0.29 c16:3n-6 0.46 ± 0.03 0.58 ± 0.03 0.29 ± 0.05 0.26 ± 0.05 0.30 ± 0.03 c16:4n-1 0.52 ± 0.05 0.58 ± 0.05 0.44 ± 0.22 0.17 ± 0.06 0.20 ± 0.03

c17:0 0.67 ± 0.07 0.37 ± 0.06 0.29 ± 0.04 0.17 ± 0.01 0.33 ± 0.03 c18:0 1.92 ± 0.07 1.24 ± 0.04 3.24 ± 0.07 2.31 ± 0.15 4.15 ± 0.15

c18:1n-11 0.61 ± 0.03 0.14 ± 0.00 2.09 ± 0.98 1.07 ± 0.10 1.16 ± 0.12 c18:1n-9 9.15 ± 0.41 8.23 ± 0.14 10.89 ± 0.63 13.27 ± 1.19 12.99 ± 0.40 c18:1n-7 4.05 ± 0.15 4.18 ± 0.08 4.15 ± 0.17 3.76 ± 0.15 5.53 ± 0.17 c18:1n-5 0.49 ± 0.01 0.51 ± 0.01 0.53 ± 0.01 0.48 ± 0.03 0.46 ± 0.02 c18:2n-6 0.70 ± 0.03 0.81 ± 0.03 0.75 ± 0.03 0.90 ± 0.06 0.87 ± 0.05 c18:3n-3 0.25 ± 0.01 0.26 ± 0.01 0.26 ± 0.05 0.42 ± 0.04 0.39 ± 0.05 c18:4n-3 0.68 ± 0.05 0.93 ± 0.05 0.68 ± 0.36 0.81 ± 0.03 0.41 ± 0.04

c20:1n-11 0.51 ± 0.02 0.61 ± 0.03 1.36 ± 0.29 1.40 ± 0.12 0.75 ± 0.06 c20:1n-9 8.42 ± 0.57 12.49 ± 0.49 6.97 ± 0.86 8.75 ± 0.80 4.15 ± 0.31 c20:1n-7 0.66 ± 0.04 0.89 ± 0.03 1.02 ± 0.22 0.64 ± 0.03 0.99 ± 0.13 c20:2n-6 0.12 ± 0.00 0.24 ± 0.01 0.31 ± 0.04 0.24 ± 0.02 0.41 ± 0.04 c20:4n-6 0.52 ± 0.03 0.46 ± 0.02 1.80 ± 0.38 0.84 ± 0.07 3.04 ± 0.27 c20:5n-3 11.94 ± 0.33 10.09 ± 0.28 12.10 ± 1.68 8.23 ± 0.68 13.54 ± 0.37

c22:1n-11 9.10 ± 0.66 15.46 ± 0.50 2.79 ± 0.24 7.47 ± 0.54 1.93 ± 0.22 c22:1n-9 0.95 ± 0.07 1.46 ± 0.06 0.74 ± 0.06 0.98 ± 0.06 0.39 ± 0.03 c22:5n-3 1.05 ± 0.04 0.39 ± 0.01 1.67 ± 0.02 0.80 ± 0.04 2.21 ± 0.13 c22:6n-3 15.82 ± 0.87 8.94 ± 0.31 17.94 ± 0.15 16.42 ± 1.35 15.31 ± 0.88 c24:1n-9 0.92 ± 0.04 0.45 ± 0.01 1.08 ± 0.18 0.83 ± 0.09 0.95 ± 0.10

∑ 95.75 95.21 93.58 95.22 92.92 length (cm) 13.64 ± 0.10 4.05 ± 0.08 28.85 ± 2.95 29.18 ± 0.59 19.44 ± 0.77 weight (g) 17.37 ± 0.54 0.42 ± 0.01 216.32 ± 44.02 167.39 ± 15.55 87.38 ± 11.63 lipid (%) 3.67 ± 0.30 3.10 ± 0.15 1.22 ± 0.16 3.41 ± 0.64 2.20 ± 0.18

71

Table 2.2. (cont’d) Common name Shorthorn sculpin Hagfish Polychaete worm Marlin-spike Rainbow smelt Species group Myoxocephalus scorpius Myxine glutinosa Nereis virens Nezumia bairdi Osmerus mordax

n 8 12 12 6 127 c14:0 2.07 ± 0.33 4.43 ± 0.10 2.88 ± 0.37 2.03 ± 0.10 3.09 ± 0.09 c16:0 13.90 ± 0.36 8.24 ± 0.20 17.85 ± 0.85 12.24 ± 1.05 15.60 ± 0.19

c16:1n-7 10.01 ± 0.77 9.11 ± 0.13 2.94 ± 0.17 6.46 ± 0.39 8.04 ± 0.27 c16:3n-6 0.43 ± 0.04 0.31 ± 0.01 0.49 ± 0.13 0.23 ± 0.01 0.52 ± 0.03 c16:4n-1 0.49 ± 0.06 0.19 ± 0.02 0.28 ± 0.02 0.12 ± 0.02 0.64 ± 0.04

c17:0 0.54 ± 0.08 0.07 ± 0.00 0.65 ± 0.11 0.15 ± 0.04 0.28 ± 0.02 c18:0 2.99 ± 0.15 1.59 ± 0.07 4.56 ± 0.41 2.11 ± 0.23 2.83 ± 0.08

c18:1n-11 0.53 ± 0.21 1.60 ± 0.11 0.19 ± 0.04 5.88 ± 0.97 0.84 ± 0.07 c18:1n-9 15.08 ± 1.41 21.89 ± 0.59 4.08 ± 0.28 15.84 ± 2.15 14.32 ± 0.45 c18:1n-7 7.20 ± 0.47 3.64 ± 0.13 3.42 ± 0.27 4.22 ± 0.42 3.96 ± 0.07 c18:1n-5 0.38 ± 0.03 0.37 ± 0.01 0.43 ± 0.05 0.57 ± 0.03 0.34 ± 0.01 c18:2n-6 1.06 ± 0.14 0.78 ± 0.07 1.78 ± 0.17 0.46 ± 0.02 1.03 ± 0.04 c18:3n-3 0.44 ± 0.14 0.26 ± 0.01 1.61 ± 0.13 0.16 ± 0.01 0.48 ± 0.02 c18:4n-3 0.73 ± 0.08 0.48 ± 0.03 0.33 ± 0.05 0.26 ± 0.02 0.88 ± 0.05

c20:1n-11 0.67 ± 0.09 3.00 ± 0.19 4.42 ± 0.12 2.81 ± 0.28 0.96 ± 0.05 c20:1n-9 2.27 ± 0.70 12.40 ± 0.29 1.93 ± 0.07 14.36 ± 1.76 3.95 ± 0.33 c20:1n-7 0.97 ± 0.11 1.05 ± 0.03 1.17 ± 0.17 1.31 ± 0.11 0.77 ± 0.03 c20:2n-6 0.40 ± 0.12 0.17 ± 0.01 3.86 ± 0.35 0.21 ± 0.01 0.37 ± 0.02 c20:4n-6 2.37 ± 0.32 0.39 ± 0.02 1.91 ± 0.32 0.58 ± 0.09 1.90 ± 0.09 c20:5n-3 15.56 ± 1.18 3.79 ± 0.33 17.43 ± 1.36 5.76 ± 0.29 12.84 ± 0.23

c22:1n-11 0.92 ± 0.42 13.18 ± 0.45 0.48 ± 0.07 9.37 ± 0.92 2.50 ± 0.24 c22:1n-9 0.24 ± 0.05 1.77 ± 0.05 0.19 ± 0.04 1.78 ± 0.25 0.40 ± 0.03 c22:5n-3 1.74 ± 0.22 3.61 ± 0.13 0.27 ± 0.25 0.84 ± 0.08 1.80 ± 0.08 c22:6n-3 11.44 ± 1.04 1.86 ± 0.18 3.79 ± 0.46 6.73 ± 0.51 14.75 ± 0.54 c24:1n-9 0.35 ± 0.03 0.50 ± 0.02 0.50 ± 0.07 0.50 ± 0.04 0.66 ± 0.04

∑ 92.78 94.66 77.41 94.99 9.83 93.75 3.11 length (cm) 21.55 ± 1.46 40.70 ± 0.99 20.98 ± 1.53 22.68 ± 1.08 14.97 ± 0.24 weight (g) 215.34 ± 83.96 79.41 ± 7.65 4.68 ± 0.39 35.34 ± 5.74 27.39 ± 1.64 lipid (%) 2.11 ± 0.23 15.23 ± 1.19 2.08 ± 0.33 7.51 ± 1.93 2.97 ± 0.17

72

Table 2.2. (cont’d) Common name Northern shrimp Striped shrimp Lampreys Winter flounder Pollock Species group Pandalus borealis Pandalus montagui Petromyzontidae Pseudopleuronectes americanus Pollachius virens

n 12 18 3 43 6 c14:0 2.67 ± 0.12 2.44 ± 0.23 18.68 ± 1.97 2.21 ± 0.09 2.82 ± 0.27 c16:0 14.01 ± 0.54 15.31 ± 0.28 18.60 ± 0.60 16.23 ± 0.32 14.88 ± 0.53

c16:1n-7 10.33 ± 0.53 8.53 ± 0.50 23.57 ± 0.64 7.69 ± 0.55 6.35 ± 0.50 c16:3n-6 0.50 ± 0.04 0.41 ± 0.06 0.07 ± 0.01 0.17 ± 0.01 0.41 ± 0.05 c16:4n-1 0.22 ± 0.04 0.27 ± 0.08 0.20 ± 0.07 0.10 ± 0.02 0.52 ± 0.07

c17:0 0.17 ± 0.01 0.37 ± 0.03 0.12 ± 0.03 0.43 ± 0.02 0.56 ± 0.17 c18:0 1.87 ± 0.13 2.36 ± 0.10 1.78 ± 0.30 3.68 ± 0.13 3.72 ± 0.45

c18:1n-11 0.58 ± 0.10 0.30 ± 0.04 0.41 ± 0.29 0.42 ± 0.05 1.22 ± 0.27 c18:1n-9 13.23 ± 0.44 11.86 ± 0.35 13.82 ± 0.91 9.14 ± 0.36 13.70 ± 1.79 c18:1n-7 6.27 ± 0.23 7.21 ± 0.27 2.06 ± 0.16 4.40 ± 0.14 4.43 ± 0.56 c18:1n-5 0.51 ± 0.02 0.59 ± 0.03 0.21 ± 0.01 0.48 ± 0.04 0.44 ± 0.07 c18:2n-6 0.79 ± 0.03 0.79 ± 0.04 0.35 ± 0.04 1.01 ± 0.07 0.90 ± 0.06 c18:3n-3 0.25 ± 0.01 0.23 ± 0.02 0.08 ± 0.00 0.80 ± 0.11 0.47 ± 0.03 c18:4n-3 0.53 ± 0.07 0.41 ± 0.08 0.08 ± 0.03 0.42 ± 0.04 1.18 ± 0.09

c20:1n-11 1.07 ± 0.10 0.95 ± 0.07 0.61 ± 0.09 1.16 ± 0.09 1.25 ± 0.26 c20:1n-9 5.45 ± 0.56 3.45 ± 0.62 0.69 ± 0.50 2.21 ± 0.19 6.47 ± 1.28 c20:1n-7 1.03 ± 0.06 1.39 ± 0.11 0.09 ± 0.04 1.35 ± 0.08 0.56 ± 0.05 c20:2n-6 0.31 ± 0.01 0.32 ± 0.02 0.27 ± 0.01 0.72 ± 0.07 0.29 ± 0.02 c20:4n-6 1.21 ± 0.08 1.87 ± 0.18 1.21 ± 0.30 3.83 ± 0.36 0.99 ± 0.15 c20:5n-3 15.03 ± 0.63 17.13 ± 0.72 2.81 ± 1.44 14.87 ± 0.46 12.92 ± 1.08

c22:1n-11 6.52 ± 0.77 4.69 ± 1.12 0.21 ± 0.11 0.73 ± 0.16 4.39 ± 1.09 c22:1n-9 0.92 ± 0.10 0.51 ± 0.06 0.04 ± 0.01 0.30 ± 0.04 0.65 ± 0.11 c22:5n-3 0.60 ± 0.07 0.91 ± 0.06 1.56 ± 0.08 4.79 ± 0.25 1.00 ± 0.12 c22:6n-3 10.58 ± 0.35 10.21 ± 0.51 4.26 ± 1.57 11.59 ± 0.93 14.21 ± 2.37 c24:1n-9 0.40 ± 0.03 0.43 ± 0.03 0.07 ± 0.03 0.57 ± 0.05 0.53 ± 0.09

∑ 95.03 92.93 91.82 89.28 94.86 length (cm) 10.13 ± 0.70 - - - 25.50 ± 2.02 21.17 ± 0.64 31.83 ± 2.14 weight (g) 6.99 ± 1.05 3.07 ± 0.25 42.07 ± 1.49 208.29 ± 20.56 403.02 ± 64.53 lipid (%) 2.03 ± 0.14 2.01 ± 0.12 9.83 ± 3.08 2.01 ± 0.13 4.64 ± 1.10

73

Table 2.2. (cont’d) Common name Greenland halibut (turbot) Salmon-smolts Salmon (large) Scampi shrimp Deepwater redfish Species group Reinhardtius hippoglossoides Salmo salar Salmo salar-L Sclerocrangon boreas Sebastes sp.

n 26 12 3 6 12 c14:0 4.72 ± 0.20 1.78 ± 0.09 3.82 ± 0.38 1.54 ± 0.13 3.34 ± 0.16 c16:0 16.06 ± 0.49 16.80 ± 0.23 14.47 ± 1.20 14.30 ± 0.20 14.08 ± 0.53

c16:1n-7 9.58 ± 0.25 8.37 ± 0.68 7.10 ± 0.57 7.76 ± 0.95 6.75 ± 0.41 c16:3n-6 0.55 ± 0.04 0.66 ± 0.06 0.40 ± 0.12 0.28 ± 0.04 0.50 ± 0.07 c16:4n-1 0.30 ± 0.02 0.26 ± 0.02 0.27 ± 0.10 0.22 ± 0.03 0.33 ± 0.05

c17:0 1.11 ± 0.12 0.42 ± 0.02 0.24 ± 0.06 0.61 ± 0.06 0.13 ± 0.01 c18:0 2.70 ± 0.09 5.33 ± 0.15 2.52 ± 0.07 2.56 ± 0.21 3.64 ± 0.11

c18:1n-11 0.54 ± 0.05 0.01 ± 0.00 0.71 ± 0.19 0.11 ± 0.01 0.95 ± 0.12 c18:1n-9 13.72 ± 0.56 11.53 ± 0.60 14.59 ± 0.54 10.03 ± 0.78 11.92 ± 0.56 c18:1n-7 5.39 ± 0.19 5.43 ± 0.28 4.10 ± 0.41 9.44 ± 0.29 5.45 ± 0.32 c18:1n-5 0.40 ± 0.02 0.16 ± 0.02 0.47 ± 0.01 0.92 ± 0.09 0.46 ± 0.02 c18:2n-6 0.69 ± 0.03 3.70 ± 0.27 0.92 ± 0.15 0.91 ± 0.03 0.88 ± 0.03 c18:3n-3 0.22 ± 0.01 5.13 ± 0.36 0.41 ± 0.06 0.20 ± 0.01 0.21 ± 0.01 c18:4n-3 0.60 ± 0.04 1.14 ± 0.08 0.79 ± 0.04 0.42 ± 0.07 0.48 ± 0.06 c20:1n-11 0.86 ± 0.11 0.03 ± 0.01 1.11 ± 0.30 1.19 ± 0.13 1.10 ± 0.07 c20:1n-9 8.53 ± 0.70 0.40 ± 0.03 8.83 ± 1.03 0.79 ± 0.09 9.26 ± 0.81 c20:1n-7 1.19 ± 0.12 0.15 ± 0.01 0.60 ± 0.02 1.64 ± 0.19 0.90 ± 0.04 c20:2n-6 0.26 ± 0.02 0.29 ± 0.02 0.26 ± 0.04 0.37 ± 0.02 0.19 ± 0.01 c20:4n-6 0.70 ± 0.17 2.36 ± 0.18 0.36 ± 0.03 2.72 ± 0.32 1.24 ± 0.10 c20:5n-3 7.78 ± 0.60 10.27 ± 0.55 8.47 ± 1.05 20.76 ± 0.70 9.00 ± 0.43 c22:1n-11 8.19 ± 0.79 0.02 ± 0.00 8.92 ± 1.22 0.38 ± 0.03 9.75 ± 0.87 c22:1n-9 1.12 ± 0.09 0.06 ± 0.01 1.14 ± 0.15 0.24 ± 0.03 1.41 ± 0.11 c22:5n-3 1.22 ± 0.19 4.01 ± 0.17 2.33 ± 0.47 1.83 ± 0.14 0.59 ± 0.03 c22:6n-3 7.30 ± 0.48 14.67 ± 1.20 10.93 ± 0.74 11.03 ± 0.95 12.04 ± 0.78 c24:1n-9 0.59 ± 0.04 0.39 ± 0.06 0.77 ± 0.05 0.22 ± 0.02 0.70 ± 0.05

∑ 94.31 93.36 94.51 90.45 95.32 length (cm) 25.55 ± 1.31 14.45 ± 0.25 54.97 ± 1.77 9.10 ± 0.49 16.83 ± 0.62 weight (g) 166.39 ± 23.85 26.98 ± 1.40 2285.70 ± 183.46 17.89 ± 2.59 116.74 ± 12.55 lipid (%) 5.55 ± 0.68 2.26 ± 0.26 12.62 ± 1.31 1.44 ± 0.25 2.17 ± 0.21

74

Table 2.2. (cont’d) Common name Bobtail squid Hyperiid amphipod Krill Moustache sculpin Longfin hake Species group Semirossia tenera Themisto libellula Thysanoessa raschii Triglops murrayi Urophycis chesteri

n 7 16 5 9 6 c14:0 2.65 ± 0.26 4.46 ± 0.14 4.80 ± 0.10 3.14 ± 0.15 3.31 ± 0.16 c16:0 14.71 ± 1.11 10.10 ± 0.59 15.29 ± 0.27 16.14 ± 0.29 10.38 ± 0.51

c16:1n-7 4.11 ± 0.91 7.91 ± 0.28 9.76 ± 0.16 11.00 ± 0.44 7.75 ± 0.27 c16:3n-6 0.09 ± 0.02 0.43 ± 0.02 1.41 ± 0.06 0.79 ± 0.05 0.46 ± 0.03 c16:4n-1 0.10 ± 0.01 0.30 ± 0.05 3.26 ± 0.20 0.77 ± 0.06 0.27 ± 0.05

c17:0 0.49 ± 0.09 0.20 ± 0.03 1.92 ± 0.13 0.83 ± 0.05 0.08 ± 0.01 c18:0 3.16 ± 0.26 0.75 ± 0.06 2.27 ± 0.07 2.68 ± 0.10 1.58 ± 0.13

c18:1n-11 0.69 ± 0.10 0.40 ± 0.03 0.08 ± 0.01 0.41 ± 0.03 5.98 ± 0.73 c18:1n-9 6.52 ± 1.00 10.21 ± 0.63 8.78 ± 0.11 15.54 ± 0.67 11.98 ± 1.56 c18:1n-7 3.80 ± 0.17 3.32 ± 0.25 4.91 ± 0.18 7.64 ± 0.12 5.08 ± 0.37 c18:1n-5 0.60 ± 0.02 0.69 ± 0.03 0.26 ± 0.01 0.35 ± 0.02 0.49 ± 0.02 c18:2n-6 0.64 ± 0.03 0.78 ± 0.03 0.55 ± 0.02 0.85 ± 0.03 0.57 ± 0.03 c18:3n-3 0.22 ± 0.03 0.29 ± 0.02 0.16 ± 0.01 0.30 ± 0.02 0.20 ± 0.01 c18:4n-3 0.30 ± 0.05 0.92 ± 0.11 2.13 ± 0.08 1.18 ± 0.07 0.49 ± 0.04

c20:1n-11 1.48 ± 0.25 3.91 ± 0.37 0.41 ± 0.03 0.36 ± 0.02 3.66 ± 0.70 c20:1n-9 8.86 ± 0.75 15.86 ± 1.14 7.78 ± 0.35 2.08 ± 0.16 14.49 ± 1.55 c20:1n-7 0.96 ± 0.17 1.14 ± 0.04 0.81 ± 0.05 0.87 ± 0.07 1.30 ± 0.07 c20:2n-6 0.57 ± 0.03 0.19 ± 0.01 0.15 ± 0.00 0.18 ± 0.01 0.15 ± 0.01 c20:4n-6 1.75 ± 0.43 0.27 ± 0.01 0.33 ± 0.01 0.89 ± 0.06 0.35 ± 0.05 c20:5n-3 16.53 ± 0.98 8.20 ± 0.58 14.62 ± 0.24 15.57 ± 0.34 5.33 ± 0.37

c22:1n-11 5.14 ± 1.02 15.61 ± 1.42 6.37 ± 0.50 1.06 ± 0.12 12.29 ± 1.63 c22:1n-9 1.20 ± 0.21 1.83 ± 0.14 0.97 ± 0.06 0.24 ± 0.01 1.97 ± 0.21 c22:5n-3 0.70 ± 0.10 0.37 ± 0.02 0.46 ± 0.01 0.94 ± 0.03 0.83 ± 0.05 c22:6n-3 16.73 ± 1.39 7.02 ± 0.34 5.96 ± 0.15 9.31 ± 0.73 6.00 ± 0.76 c24:1n-9 0.50 ± 0.04 0.41 ± 0.02 0.28 ± 0.02 0.36 ± 0.04 0.47 ± 0.03

∑ 92.50 95.55 93.68 93.47 95.44 length (cm) 4.74 ± 0.69 - - - - - - 10.41 ± 0.42 24.67 ± 1.77 weight (g) 12.17 ± 4.26 0.36 ± 0.03 0.47 ± 0.03 8.81 ± 1.24 162.96 ± 41.44 lipid (%) 1.81 ± 0.26 4.82 ± 0.51 6.47 ± 0.38 3.96 ± 0.32 8.85 ± 1.33

75

Table 2.2. (cont’d) Common name White hake Ocean pout Species group Urophycis tenuis Zoarces americanus

n 7 1 c14:0 2.16 ± 0.09 2.15 c16:0 13.53 ± 0.13 14.09

c16:1n-7 5.26 ± 0.38 11.44 c16:3n-6 0.20 ± 0.02 0.37 c16:4n-1 0.11 ± 0.02 0.27

c17:0 0.28 ± 0.02 0.15 c18:0 4.07 ± 0.15 3.02

c18:1n-11 2.95 ± 0.16 0.92 c18:1n-9 11.74 ± 0.50 20.65 c18:1n-7 5.39 ± 0.24 5.91 c18:1n-5 0.49 ± 0.03 0.43 c18:2n-6 0.71 ± 0.03 0.81 c18:3n-3 0.23 ± 0.02 0.22 c18:4n-3 0.31 ± 0.03 0.38 c20:1n-11 1.47 ± 0.06 0.98 c20:1n-9 7.76 ± 0.42 3.76 c20:1n-7 0.83 ± 0.06 1.59 c20:2n-6 0.25 ± 0.01 0.14 c20:4n-6 1.67 ± 0.12 2.44 c20:5n-3 9.14 ± 0.29 9.68 c22:1n-11 3.82 ± 0.28 1.26 c22:1n-9 0.73 ± 0.03 0.31 c22:5n-3 1.70 ± 0.04 0.91 c22:6n-3 19.38 ± 0.93 11.17 c24:1n-9 0.72 ± 0.12 0.45

∑ 94.88 93.45 length (cm) 26.17 ± 3.91 29.90 weight (g) 256.60 ± 184.97 125.51 lipid (%) 1.92 ± 0.19 1.83

Table 2.3. Principal component (PC) analysis using all specimens (n = 1028) and the set

of 16 logratio fatty acids. Summary of significant (>|0.5|) factor loadings for

the retained PCs (eigenvalues >1) following Varimax rotation. Maximum

absolute loadings on each fatty acid are indicated in bold.

Varimax-rotated factor

Logratio FA PC 1 PC 2 PC 3 PC 4 14:0 0.53 0.58 16:0 0.61

16:1-7 0.71 16:4-1 -0.51 0.59 18:1-9 0.67 18:1-7 0.57 18:2-6 0.85 18:3-3 0.85 18:4-3 0.56 0.58 20:1-11 0.87 20:1-9 0.70 20:4-6 -0.90 20:5-3 -0.79 22:1-11 0.51 0.69 22:1-9 0.70 22:6-3 -0.84

% of variance 17.66 25.17 17.91 20.92

77 Table 2.4. Principal component (PC) analysis using eight prey species commonly found

in the Estuary in autumn and the set of 16 logratio fatty acids. Summary of

significant (>|0.5|) factor loadings for the retained PCs (eigenvalues >1)

following Varimax rotation. Maximum absolute loadings on each fatty acid

are indicated in bold.

Varimax-rotated factorLogratio FA PC 1 PC 2

14:0 0.74 0.63 16:0 0.72 0.65

16:1-7 0.61 0.73 16:4-1 0.80 18:1-9 0.87 18:1-7 0.58 0.59 18:2-6 0.89 18:3-3 0.88 18:4-3 0.82 20:1-11 0.69 20:1-9 0.90 20:4-6 -0.75 20:5-3 0.79 22:1-11 0.97 22:1-9 0.93 22:6-3 0.83

% of variance 50.55 29.06

78

Figure 2.1. a) Study area (box) and b) sampling zones in the Estuary and Gulf of St.

Lawrence, Canada.

79

Figure 2.2. Selected species with significant amounts of the nominally ‘trace-level’ fatty

acid 22:1n-7 not used in further evaluations. Dashed line = global mean of

0.27%. Diamonds are the horizontal spread across the mean (middle lines),

with significant differences between species indicated by non-overlapping

vertical apices.

80

Figure 2.3. Cumulative mean abundance of four minor (global mean contribution

<0.5%) polyunsaturated fatty acids. Species groups are ranked in descending

order of abundance of the fatty acid 18:2n-6.

81

Figure 2.4. Matrix scatterplot of factor scores on the retained PC Factors 1 through 4,

accounting for 81.7% of total cumulative variance in the 16 logratio fatty

acid variables on all samples. Groups labelled are among those that scored

farthest from the central majority.

82

a)

b)

Figure 2.5. Scatterplot of scores on Factor 1 and 2 retained from a separate PC analysis

conducted on eight species of prey commonly found in the Estuary in

autumn. Individual scores are indicated by species name (a) and lipid

content (b).

83

Figure 2.6. Original PC analysis showing only the scores for four cod species.

84

Figure 2.7. Original PC analysis showing only the scores for smelt, capelin, and herring.

Overlapping scores on the first two PC factors (enlarged, in upper right, with

convex area shading according to species) are in contrast to separations seen

in the full matrix with Factors 3 and 4 (lower left).

85

Figure 2.8. Example of within-species spatiotemporal variation, with plotted scores on

Factors 1 and 2 for all capelin specimens by year, season, and locations: SI =

Sept-Iles (N. Gulf), NS = North Shore (Lower Estuary), SS = South Shore

(Lower Estuary), UE = Upper Estuary.

86

Figure 2.9. Example of between-species seasonal variability, with plotted scores on

Factors 1 and 2 for autumn specimens of smelt (triangles), capelin

(diamonds), herring and sardines (crosses).

.

87

Figure 2.10. Scores on Factors 1 and 2 for autumn samples weighted by total lipid content

for smelt (orange), capelin (green) and herring (blue) and herring sardines

(cyan).

88

Figure 2.11. Biplot on the first two discriminant functions from an analysis classifying

those prey species sampled in greatest number (range = 29–126), comprising

three pelagic fishes (smelt, capelin, herring), three demersal fishes (flounder,

plaice, tomcod), and one invertebrate (crangon shrimp). Circles represent

95% confidence intervals of mean scores by species. Also projected are the

more-important contributions from 11 of the 16 logratio fatty acids used in

the classification along with individual scores.

89

Figure 2.12. Biplot on the first two discriminant functions from an analysis classifying 15

species groups (>17 samples each, total n = 699) of potential prey of beluga,

using 16 fatty acid logratios as input variables. Circles denote 95%

confidence intervals for mean scores by group (sculpins and shrimps

comprised three species each). Non-overlapping circles indicate significantly

different groups. Oblique labels suggest species associations. For reference,

90

scores are plotted for all samples, including those not used in the analysis,

such as eels and whelks (shaded areas). Selected individuals of Greenland

halibut are marked as being distinct from their species group. Variables are

projected according to their relative contributions on the two discriminant

functions. For clarity, the minor contributions of fatty acids 18:3n-3, 20:1n-

11, and 22:1n-9 are not included.

91

Chapitre 3

Comparing fatty acid signatures among marine

mammals of the Estuary and Gulf of St. Lawrence: an

evaluation using multivariate methods.

92

Résumé

Les profils en acides gras du lard chez les bélugas (Delphinapterus leucas) ainsi que

chez les phoques communs (Phoca vitulina), les phoques gris (Halichoerus grypus), les

phoques du Groenland (Pagophilus groenlandicus) et les phoques à capuchon (Cystophora

cristata) ont été examinés à l’aide de méthodes d’analyse exploratoire des données. Les

méthodes multivariées utilisées incluaient les arbres de classification des analyses en

composantes principales, de groupements et de fonctions discriminantes. Les différences et

les similarités ainsi démontrées avec 17 acides gras relativement abondants et liés à

l’alimentation ont suggéré que la discrimination s’effectue entre les espèces, et, moins

fortement, entre les classes incluant le sexe et l’âge. Les acides gras polyinsaturés 20:5n-3

et 22:6n-3 étaient d’intérêt spécial pour classifier le béluga à part des pinnipèdes, et les

pinnipèdes juvéniles à part des adultes. La réussite de la classification souligne le potentiel

des profils en acides gras pour dévoiler l’histoire alimentaire des résidents et des visiteurs

de l’estuaire et du golfe du Saint-Laurent, Canada.

Abstract

Fatty acid signatures of blubber in beluga (Delphinapterus leucas) along with

harbour seals (Phoca vitulina), grey seals (Halichoerus grypus), harp seals (Pagophilus

groenlandicus), and hooded seals (Cystophora cristata) were examined using exploratory

data analysis. The multivariate methods used were classification trees, along with a suite of

principal component analysis, cluster analysis, and discriminant function analysis.

Differences and similarities using 17 diet-linked and relatively-abundant fatty acids

suggested discrimination could be achieved between these species, and to lesser extent

93

within classes including gender and age. Marine-linked polyunsaturated fatty acids 20:5n-3

and 22:6n-3 were of special interest in classifying belugas from pinnipeds and subadult

pinnipeds from adults. The success at classification suggests the potential of fatty acid

profiles to reveal the dietary history of residents and migrants for the regions of the Estuary

and the Gulf of St. Lawrence, Canada.

94

Introduction

The Estuary and Gulf of St. Lawrence are highly productive regions of the

Northwest Atlantic (El-Sabh and Silverberg 1990, Therriault 1991). These areas also

represent the seasonal or permanent home to 15 species of marine mammals, including

several that are of special concern (Lavigueur et al. 1993, Kingsley and Reeves 1998,

COSEWIC 2003, Kinze 2003). The precarious status of certain populations is largely a

result of historical overexploitation (Mitchell 1974, Reeves and Mitchell 1987, Kingsley

2002). However, urbanisation, fisheries, and climate changes have also occurred in a

number of marine ecosystems over the last few decades, including the Estuary and Gulf of

St. Lawrence (Pauly et al. 1998a, Hammill et al. 1999, Pierce et al. 1999, Hanson and

Chouinard 2002, Zwanenburg et al. 2002, Law et al. 2003). The influence of these changes

on the structure and carrying capacity of ecosystems and the rate of recovery of precarious

species remains largely unknown (Bowen et al. 2003, Myers and Worm 2003, Trites and

Donnelly 2003, Greene and Pershing 2004, Frank et al. 2005).

An understanding of the trophic interactions among components of food webs is

necessary to appreciate the effects of ecosystem changes on species, and the role that top

predators such as marine mammals may play in these ecosystems (Bowen 1997, DeMaster

et al. 2001, Yodzis 2001, Springer et al. 2003, Whitehead et al. 2003). Diet composition

has commonly been inferred from the analysis of prey remains in digestive tracts of

predators (Pitcher 1980, Harvey and Antonelis 1994, Lawson et al. 1998a, Cherel and

Duhamel 2004). This approach provides information on ingested species, but bears several

important biases related mainly to the differential availability or erosion of hard parts of

prey during digestion (Pierce and Boyle 1991, Burns et al. 1998, Staniland 2002, Arim and

95

Naya 2003). Other weaknesses are that digestive tracts provide information only on the last

meal prior to sampling, that digestive tracts of individuals either hunted or found stranded

on beaches are frequently empty, and that their collection through harvesting should not be

considered for species with small populations. Visual observations and video records of

feeding have also been used to obtain direct information on foraging behaviour and

ingested prey items (Bowen et al. 2002, Born et al. 2003, Davis et al. 2003, Levermann et

al. 2003, Williams et al. 2004). However, the paucity of direct observations of feeding in

most species of marine mammals and the high costs of available technology to monitor

feeding underwater places limits on the use of these techniques.

The limitations of direct techniques have led to interest in alternative or

complementary approaches to the study of diet of marine predators. For example, stable

carbon and nitrogen isotope ratios have been used to examine trophic interactions in aquatic

food webs, including those of the Estuary and Gulf of St. Lawrence (Minigawa and Wada

1984, Michener and Schell 1994, Lesage et al. 2001). While this technique permits the

general examination of carbon sources and relative trophic positions of food web

components, it usually does not allow for the identification of specific prey (Gannes et al.

1998, Ben-David and Schell 2001). Other indirect approaches to understanding diet may

use indicators of prey ingestion such as changes in stomach temperature (Hedd et al. 1996)

and diving profiles (Folkow and Blix 1999), or information from other suitable markers

such as radioisotopes (Born et al. 2002), synthetic contaminants (Hobbs et al. 2003), fecal

DNA (Reed et al. 1997), natural pigments (Howell et al. 2004), and marine lipids

(Dalsgaard et al. 2003).

The method utilising lipids to investigate diet focuses on the analysis of fatty acids

contained in the fat stores of predators. Fatty acids are a component of lipids, usually

96

comprised of an even number of carbon atoms bonded as a linear chain of 14 to 24 carbons.

A wide range of longer-chain (16-22 carbon) fatty acids with one or more unsaturated

bonds have their origin in plants such as the phytoplankton in the marine environment

(Jamieson and Reid 1972, Sargent et al. 1985, Ackman 1989). In general, species from

higher trophic levels, such as fishes and mammals, are unable to produce the longer-chain,

unsaturated fatty acids and therefore must obtain these elements from their diet (Greene and

Selivonchick 1987, Sheridan 1994, Cook 1996, Nakamura and Nara 2003).

While it is assumed that diet plays an important role, the establishment of a direct

link between a prey species and a specific fatty acid detected in predator lipids is not

usually possible except in special cases. Examples of clear linkages between fatty acid

compositions and diet include specialised predators such as marine turtles feeding on

jellyfish which contain uncommon fatty acids (Holland et al. 1990), or predators that

depend on environments associated with distinctive sources of fatty acids, as is seen with

pinnipeds feeding in freshwater vs. marine locations (Smith et al. 1996, Käkelä and

Hyvärinen 1998). An alternative solution to relying on specific fatty acids for the study of

diet composition is the examination of the relative proportion of a range of fatty acids that

may be detected in tissues of predators. This suite of fatty acids may then serve as a profile

or signature that can be effective at discriminating between individual predators (Smith et

al. 1997). In addition, the use of such signature information has been shown to potentially

reveal qualitative and quantitative information on diets when employed in a powerful

modelling exercise, or QFASA (Iverson et al. 2004). The examination of fatty acid

signatures among sympatric predators might therefore provide insights into the feeding

habits of the different species and their variability between regions, periods, age-classes or

97

genders (Iverson et al. 1997a, Iverson et al. 1997b, Brown et al. 1999, Walton et al. 2000,

Dahl et al. 2003, Andersen et al. 2004).

In this study, the fatty acid signatures of one cetacean and four species of pinniped

that are either resident (beluga whales, Delphinapterus leucas, harbour seals, Phoca

vitulina, and some grey seals, Halichoerus grypus) or seasonal visitors (harp seals,

Pagophilus groenlandicus, and hooded seals, Cystophora cristata) to the Estuary and Gulf

of St. Lawrence were compared to gain insights into their relative degree of dietary overlap.

In order to better accommodate the large number of variables (fatty acids) relative to the

number of observations (individual marine mammals), fatty acid composition of individuals

was examined in parallel using two exploratory, multivariate statistical approaches: 1) a

series of recursive partitioning or classification trees (CART) and, 2) a combination of

principal components, cluster and discriminant functions analyses.

3.2. Materials and Methods

Field collection

Blubber samples were obtained from 143 marine mammals from the Estuary and

Gulf of St. Lawrence during 1996−1998: 19 beluga whales, 43 harbour seals, 31 grey seals,

31 harp seals, and 19 hooded seals (Table 3.1). Beluga samples were collected from

carcasses beach-cast during the ice-free period in the lower St. Lawrence Estuary (Fig. 3.1).

Samples from pinnipeds were obtained either through biopsy sampling of live-captured

individuals or from sampling animals shot during scientific collections. Handling

procedures for live-captured animals are presented in detail in Lesage et al. (2001). The

pinnipeds were comprised of breeding (adult) harp, hooded, and grey seals sampled in

98

whelping patches in the Gulf of St. Lawrence during winter. Harbour seals, including

adults (2+ years), sub-adults (yearlings) and neonates (pups), along with non-breeding adult

(2+) and sub-adult (yearling) grey and harp seals were obtained from the lower St.

Lawrence Estuary. All blubber samples were wrapped in aluminium foil and kept frozen in

plastic bags at -20°C until processed.

Chemical analyses

For pinnipeds, the entire depth of the blubber layer from skin to muscle was used for

fatty acid analyses. Owing to the more pronounced stratification of fatty acids in the

blubber of toothed whales, the blubber layer of belugas was divided into tiers, and only

analyses of the inner-most layer were retained (Worthy and Abend 1997, Koopman 2001,

Krahn et al. 2004). Lipids were extracted from homogenised blubber tissue samples and

esterified to fatty acid methyl esters (FAME) using a modified Folch method (Iverson et al.

2002). Briefly, tissue aliquots of approximately 1.5 g were solvent extracted with 2:1

chloroform-methanol. The lower lipid fraction was filtered and evaporated under nitrogen.

The prepared FAME were analysed in duplicate on a Perkin-Elmer Autosystem II capillary

gas chromatograph with flame ionisation detector and a flexible fused coated silica column

(30 x 0.25 mm) (Iverson et al. 1997b). Fatty acids are presented here as mean mass

percent of total fatty acids, and are designated using the standard notation X:Yn-Z, where X

indicates carbon chain length, Y represents the number of double bonds and Z represents

the position of the double bond nearest to the terminal methyl group.

99

Statistical analyses

Over 50 fatty acids and their isomers are routinely identified in quantities exceeding

detection levels. Taking the smallest sample size for a species, minus one, a set of 18 fatty

acids were selected for statistical analyses on the basis of their abundance and for their

potential in serving as diet indicators (Iverson 1993, Kirsch et al. 2000).

Fatty acid data were logratio transformed before statistical treatment, as is

recommended for compositional data (Aitchison 1986, Budge et al. 2002). Values were

transformed by first normalising over 100% the proportions of the 18 selected fatty acids,

then calculating the logarithm of the ratio of a given fatty acid to that of the fatty acid 18:0.

The fatty acid 18:0 was selected as the standard as it is usually present in small but stable

amounts (Koopman 2001, Budge et al. 2002). The transformation resulted in 17 logratio

variables for use in statistical comparisons.

Variability in fatty acid composition between species, age classes or gender was

investigated using two multivariate approaches. Fatty acid compositions were initially

examined using a recursive partitioning technique, more commonly referred to as

classification and regression trees (CART) (Venables and Ripley 1999). CART has the

benefit that it does not require the transformation of data to meet normality assumptions,

nor that the number of variables be several times lower than sample size, thus permitting

the use of the original fatty acids instead of logratios (Breiman et al. 1984, Hastie et al.

2001). CART operates by minimising within-group deviations while maximising between-

group variance through successive binary splits using one variable at a time (Breiman et al.

1984, Bakshi and Utojo 1999). CART achieves this task by dividing cases into tree nodes,

or groups, that increase class homogeneity. Splitting terminates once a node contains a

100

pure class, or when no further splits can be made to improve class purity (Breiman et al.

1984, Venables and Ripley 1999). Results are presented in the form of a binary decision

tree, with successive nodes indicating the value (% of mass total) of the fatty acid variable

used as the splitting criterion, and terminal nodes, or leaves, representing the group

composition, such as the proportion of a species, age-class or gender included in the node.

In seeking to split all cases into homogenous groups, the resulting overgrown trees may

become difficult to interpret. In order to minimise this effect, the tree model may then be

pruned back to higher nodes to achieve an optimal tree size, which is characterised by a

high purity score (R2 of 1 = pure; 0 being poor) and relatively few branches (Breiman et al.

1984, Bakshi and Utojo 1999, Venables and Ripley 1999).

An initial CART analysis was produced using all 56 fatty acids. Additional

analyses were made while using the reduced set of 18 fatty acids. CART models that had

splits based on fatty acids in small amounts (<1%) were alternately investigated by

excluding the minor fatty acid, thus inducing CART to select the next-best fatty acid for

splitting (based on purity score), which may be a fatty acid found in greater amounts, and

thus potentially easier to evaluate than those splits decided on trace amounts. The

classification success of the CART models was evaluated using the number of misclassified

individuals by class in each of the terminal nodes (Smith et al. 1997).

Fatty acid signatures of marine mammals were also examined using a combination

of principal component and cluster analysis, followed by validation using discriminant

function analysis. Initially, the set of 17 logratio fatty acids was introduced into a principal

component analysis (PCA). This operation reduced the large number of correlated fatty

acid variables into a smaller set of orthogonal factors while retaining most of the original

variance (Hair et al. 1998). Factor rotation (Varimax procedure) further facilitated

101

interpretation of the results by emphasising the correlations of variables with a given factor

(Hair et al. 1998). The retained factors (eigenvalues > 1) and their corresponding scores for

each observation of an individual beluga or seal were then used in a hierarchical

agglomerative cluster analysis with Ward’s method of linkage (Martin-Fernandez and

Barcelo-Vidal 1998). The classification success of the composite PC factors and derived

cluster solutions were cross-validated using linear discriminant function analysis (Legendre

and Legendre 1998, Field 2000). A second series of discriminant function analyses was

conducted while using the 17 logratio fatty acids directly as input variables to evaluate the

classification success rate when not performing a factor reduction.

All statistical analyses were performed using JMP® 5.1 (SAS Institute Inc, Cary,

NC) and SPSS® 11.0 (SPSS Inc., Chicago, IL) software packages. Standard error (SEM) is

reported as a measure of variability of the mean. The alpha level for statistical significance

was set at 0.05.

3.3. Results

General patterns in the abundance of the different fatty acids

Approximately 56 fatty acids were identified and quantified in the blubber lipids of

143 individuals of five marine mammal species. Summary data are presented for 33 fatty

acids with mean overall abundances ≥ 0.2% of the total (Table 3.2). The 18 fatty acids that

were retained for statistical analyses were 14:0, 16:0, 16:1n-7, 18:0, 18:1n-7, 18:1n-9,

18:1n-11, 18:2n-6, 18:3n-3, 18:4n-3, 20:1-7, 20:1n-9, 20:1n-11 20:4n-3, 20:5n-3, 22:1n-9,

22:1n-11, 22:6n-3, which together contributed an average 86–91% (by species) of the total

mass of fatty acids.

102

Although there were some similarities in abundances for these fatty acids, large

variations in their relative quantities between marine mammal species were also readily

evident. The harbour seals, which were dominated by sub-adults, had the highest amounts

of fatty acid 16:1n-7 (19%) compared to the other species. Hooded seals had the highest

levels of fatty acids 18:1n-9 and 20:1n-9 (22 and 14%, respectively), although the latter was

also abundant in harp seals (11%). The lowest levels of 22:1n-11 were found in grey seals

and harbour seals (approx. 3%), whereas the highest levels of this fatty acid were observed

in belugas (9%). The abundance of each of the polyunsaturated fatty acids varied

significantly across species (one-way ANOVA on logratio variables; df = 4, 138; F =

27.54-132.97; all P < 0.05). Beluga whales had significantly lower amounts of each of

these polyunsaturated fatty acids (post-hoc Tukey’s HSD pairwise means comparisons, a =

0.05), but harbour seals (all age-classes) and hooded seals also had lower average amounts,

notably for the fatty acid 22:6n-3, relative to those detected in the grey seals and harp seals.

Classification models

Recursive partitioning

Classification tree analyses using 56 fatty acids produced an optimal tree model

with five terminal groups, and correctly classified 133 of the 143 individuals (R2 = 0.86) by

their species (Figure 3.2a). As the initial splitting variable, the polyunsaturated fatty acid

20:4n-3 was found in lower amounts (< 0.38%) in species that are resident to the St.

Lawrence Estuary (i.e., beluga and harbour seals) compared to species that are likely to

occur there on a seasonal basis (i.e., grey seals, harp seals and hooded seals). Harbour seals

were split from beluga whales on the basis of higher levels (≥ 0.07%) of a trace fatty acid

(18:2n-7). In the seasonal visitor branch of the tree, harp seals were split from hooded and

103

grey seals on the basis of larger amounts (≥ 6.92%) of another polyunsaturated fatty acid,

i.e., 20:5n-3, while the latter two species were split on their levels of the metabolic

intermediate fatty acid 22:5n-3.

An alternate CART analysis restricted to the subset of 18 fatty acids had 132 of 143

individuals correctly classified into five species groups (R2 = 0.82) (Figure 3.2b). Once

again, resident species to the Estuary were split from the seasonal visitors on the basis of

fatty acid 20:4n-3. In this model, instead of the trace fatty acid 18:2n-7, the more abundant

and polyunsaturated fatty acid 20:5n-3 was chosen to separate belugas and harbour seals.

This fatty acid was also responsible for the segregation of harp seals from hooded seals and

grey seals in the CART model using both the full and the reduced set of fatty acids. The

fatty acid, 20:5n-3, was the least abundant in belugas (mean =1.96%, SEM = 0.67), most

abundant in harp seals (mean = 8.73%, SEM = 2.06), and intermediate (mean = 4.15–

5.88%) in the other three phocid species. Another difference between tree models was that

diet-linked polyunsaturated fatty acid (22:6n-3), which was found in greater amounts in

grey seals (mean = 9.77%, SEM = 1.43) than hooded seals (mean = 5.96%, SEM = 0.72),

replaced the initial choice of the metabolic intermediate 22:5n-3 (excluded from the smaller

set of fatty acids) in the partitioning between hooded seals and grey seals.

The investigation of age- and gender-class within each species suggested differences

in fatty acid composition could also be revealed by CART analyses. Considering only adult

individuals, males and females of the five marine mammal species were separated on the

basis of different fatty acids and amounts, with varying degrees of success (Figure 3.3).

Summarising the tree model results by species, we found: 1) lower amounts of fatty acid

18:2n-6 and substituted 14:0 in male belugas compared to females (splits at 0.82% and

104

7.06%, R2 = 0.72 and 0.48, misclassification rate = 1/17 and 2/17, respectively); 2) lower

amounts of fatty acid 14:0 and higher amounts of substituted 18:1n-7 in male hooded seals

compared to females (splits at 3.64 and 4.64%, R2 = 0.75 and 0.59, misclassification rates =

1/19 and 2/19, respectively); 3) higher amounts of fatty acid 22:1n-11 in 2 out of 7 male

harp seals (split at 9.28%) and lower amounts of 18:2n-6 in 6 out of 8 male grey seals

(split at 1.19%) compared to females of their respective species (R2 = 0.54 and 0.55;

misclassification rate = 2/18 and 2/16, respectively); 4) lower amounts of fatty acid 18:3n-3

and higher amounts of substituted 18:1n-11 in male harbour seals compared to females

(splits at 0.38% and 5.78%, R2 = 0.31 and 0.17, misclassification rate = 2/11 and 3/11,

respectively). Fatty acid signatures were less distinct between genders when sub-adults

and adults were pooled together, as indicated by high misclassification rates in the three

species where more than one age class was sampled (grey seals = 9/31, harp seals = 8/31,

harbour seals = 18/43). When classifying sub-adults only by gender, harbour seals and grey

seals were also less distinct in their fatty acid composition (R2 = 0.16 and 0.19;

misclassification rate = 7/24 and 4/15, respectively), with greater classification success seen

for sub-adult harp seals (R2 = 0.66, misclassification rate = 1/13).

Investigation of age-class differences in fatty acid composition revealed that sub-

adult harp seals and grey seals differed to some extent from the adults of their respective

species. Sub-adult harp seals all had levels of fatty acid 22:1n-9 lower than 0.80%, similar

to amounts observed in nearly half (8/18) of the adult harp seals (Figure 3.4a). Sub-adult

grey seals had lower levels (< 0.70%) of fatty acid 20:1n-7 than adults (misclassification

rate = 1/21) (Figure 3.4b). In the case of harbour seals, adults segregated from younger

individuals on the basis of lower levels (<16.53%) of a commonly-biosynthesised fatty acid

105

(16:1n-7), whereas pups were split from sub-adults on the basis of smaller amounts (<

3.93%) of the polyunsaturated fatty acid 20:5n-3 (misclassification rate = 3/24, R2 = 0.80)

(Fig. 3.4c). The withdrawal of the fatty acid 16:1n-7 resulted in a similar tree, where

harbour seal pups segregated from older individuals on the basis of lower levels of the same

polyunsaturated fatty acid (20:5n-3), while greater amounts (≥ 5.1%) of 20:1n-7 was

selected for segregating adults from the sub-adults (misclassification rate = 4/24, R2 = 0.71).

Principal components and clustering analysis

Using a principal component analysis (PCA), the 17 logratio variables were reduced

to four factors (eigenvalues >1), which accounted for 88% of the total variance (Table 3.3).

While not shown here, the use of logratio transformed data instead of standardised fatty

acids as input variables improved the output of the PCA, with an increase in the Kaiser-

Meyer-Olkin measure of sampling adequacy from 0.62 to 0.77, and the total explained

variance from 84% to 88%, thereby confirming that the choice of transformation and

analysis was indeed appropriate for the data at hand.

The results from the PCA were also consistent with those obtained from the CART

analysis since the diet-linked polyunsaturated fatty acids highlighted for the segregation by

species in the CART analysis (i.e., 18:2n-6, 18:3n-3, 18:4n-3, 20:4n-3, 20:5n-3, 22:6n-3)

were all strongly associated with Factor 1, which explained the largest amount of variance

(37%) in the data set. Factor 2 comprised several of the diet-linked monounsaturated fatty

acids (i.e., 22:1n-9, 22:1n-11, and 20:1n-9) and explained 19% of the total variance. Factor

3 also accounted for 19% of the total variance, and was characterised by other

monounsaturated fatty acids of the 18- and 20-series. Factor 4, with 12% of the explained

variance, was associated with the saturated fatty acids 14:0 and 16:0 and the

106

monounsaturated fatty acid 16:1n-7, all of which may be commonly acquired through the

diet or produced endogenously. The abundance of the fatty acid variables associated with a

given factor all covaried positively.

Significant differences were detected between species in factor scores, and thus in

the abundance of the set of logratio fatty acids associated with each factor (one-way

ANOVA: F4,138 = 5.36-161.81; all P < 0.005) (Table 3.4). Factor 1 indicated significantly

lower levels of all polyunsaturated fatty acids in belugas than in any of the pinniped

species, with intermediate levels in harbour and hooded seals, and the highest values in harp

and grey seals (Fig. 3.5). The diet-linked monounsaturated fatty acids that were correlated

with Factor 2 were in significantly lesser amounts in harbour seals than in beluga, harp

seals or hooded seals, but not in significantly different quantities than in grey seals.

Harbour, hooded and grey seals had significantly higher levels of the fatty acids of the 18-

series and fatty acid 20:1n-11 (Factor 3) than belugas, which in turn, had intermediate but

significantly higher levels of these fatty acids than harp seals. Finally, the saturated fatty

acids 14:0 and 16:0, and fatty acid 16:1n-7 characteristic of Factor 4 were significantly less

abundant in hooded seals than in any other species, with intermediate values in grey seals,

and higher values in beluga, harp seals and harbour seals.

A discriminant function analysis as an evaluation of using PC factors in classifying

species resulted in 84% success or 120 of 143 correct assignments. The first discriminant

function emphasised the strong role of Factor 1, and therefore of polyunsaturated fatty

acids, in the segregation of belugas, and to a lesser extent harbour seals from the other

pinniped species. The second discriminant function, which represented mainly the

influence of Factor 3 with various monounsaturated fatty acids, segregated beluga and harp

seals from harbour, grey and hooded seals (Fig. 3.6a). In comparison, a discriminant

107

functions analysis conducted using the 17 fatty acid logratios directly instead of the PC

factor scores, achieved near-perfect classification success, with only 1 misclassified

individual (Fig. 3.6b).

The introduction of PCA factor scores as input variables in a hierarchical cluster

analysis joined the marine mammal individuals into nine groups (Fig. 3.7). Beluga whales,

harbour seal pups and 21 of 24 sub-adult harbour seals, all from the St. Lawrence Estuary,

segregated from adult harbour seals and from harp, hooded and grey seals to form a

separate branch of three clusters. Two of these clusters were comprised of beluga whales

only, whereas a third cluster consisted essentially of younger harbour seals and a few (3 of

18) sub-adult grey seals, also from the St. Lawrence Estuary. The other branch of the

dendrogram consisted of six clusters, of which five were mainly mono-specific in

composition and one (cluster 5) comprised a mixture of grey seals and harp seals.

The segregation of beluga whales into two clusters, one of which was dominated by

females (cluster 2) and the other by males (cluster 4), was based on significantly lower

levels of the fatty acids associated with Factors 3 and 4 (i.e., monounsaturated fatty acids of

the 18-series, fatty acids 20:1n-11 and 16:1n-7, saturated fatty acids 14:0 and 16:0) in males

than females (Table 3.7; Fig. 3.8). The occurrence of harbour seal sub-adults and pups

(cluster 7) in the same branch as beluga whales was related partly to generally low

abundances of polyunsaturated fatty acids (Factor 1) compared to the marine mammals

from the other main branch of the dendrogram, but mostly to similarities with belugas in the

abundance of the monounsaturated fatty acids of the 18-series and fatty acid 20:1n-11 (i.e.,

Factor 3), since harbour seal sub-adults and pups differed significantly from beluga whales

in all other respects (i.e., Factor 1, 2 and 4).

108

Of the 31 harp seals sampled in this study, 23 were grouped into two neighbouring

clusters (clusters 3 and 6), which were similar in all respects except in the abundance of

monounsaturated fatty acids associated with Factor 2; these monounsaturated fatty acids of

dietary origin were significantly less abundant in harp seals from cluster 3 than cluster 6 (or

5). A group of grey seals resembled hooded seals in their levels of monounsaturated fatty

acids of the 18-series and 20:1n-11 (Factor 3) and saturated fatty acids 14:0 and 16:0 and

fatty acid 16:1n-7 (Factor 4). However, these grey seals formed a separate cluster (cluster

8) from hooded seals on the basis of significantly lesser amounts of monounsaturated fatty

acids (Factor 2) and to a lesser extent, higher levels of polyunsaturated fatty acids compared

to hooded seals. The segregation of these two groups from adult harbour seals (cluster 9)

and the other grey seals and remaining eight harp seals (cluster 5) was essentially based on

their significantly lower levels of the saturated fatty acids 14:0 and 16:0, and fatty acid

16:1n-7 compared to other pinnipeds. The significantly higher abundance of

monounsaturated fatty acids of the 18-series and fatty acid 20:1n-11 in adult harbour seals

compared to the other pinnipeds, and their lower levels of polyunsaturated fatty acids

compared to grey seals and harp seals, resulted in their classification as a separate group

(cluster 9). For comparative purposes, the means (± SEM) by cluster for each fatty acid are

presented in Table 3.8.

Cross-validation of cluster memberships using a discriminant function analysis

indicated that the nine-cluster solution as presented here was optimal, having the lowest rate

of misclassifications (4/143 individuals) compared to cluster solutions with 7, 8, 10, 11 or

12 clusters. Repeating the discriminant function analysis on the nine-cluster solution while

using the original 17 logratio fatty acids as input variables produced similar results to those

109

obtained using the factor scores, even though the success rate declined slightly from 97% to

93%, with 10 individuals misclassified (Fig. 3.9).

In both sets of discriminant analyses, cross-validating results from either PCA or

cluster analyses produced similar, general patterns of projected groups, with the exception

of harbour seals and hooded seals. These two groups would switch positions when using

PCA factors as input variables, as compared to the discriminant projections with the

original logratio fatty acids. Hence, when classifying groups using either factors or

clusters, harbour seals scored closer to belugas than did the hooded seals on the first

function (Figs. 3.6a, 3.9a). Conversely, when using the set of fatty acids, both discriminant

analyses had hooded seals rather than harbour seals being projected closer to the belugas

(Figs. 3.6b, 3.9b)

3.4. Discussion

The comparative analysis of fatty acid composition of beluga whales and four

pinniped species of the Estuary and Gulf of St. Lawrence revealed differences in fatty acid

composition between species, age-classes and, to some extent, genders of these marine

mammals. However, the multivariate approaches used in this study could also vary in the

manner with which they revealed patterns between classes. Therefore, the exploratory

mechanics of the methods used will be reviewed prior to further interpretation of the

biological significance of the results.

Methodological considerations

Classification and Regression Tree (CART) and Principal Components/Cluster

analysis (hereafter called ‘PC/Cluster’) were both successful at classifying species, age-

110

classes and, to some extent, genders of the different marine mammals, and did so primarily

on the basis of the same groups of diet-linked fatty acid, usually the polyunsaturated fatty

acids. The PC/Cluster analysis also revealed some similarities in fatty acid composition

among groups from different species (e.g., adult grey seals and harp seals) as well as a

substructure within certain groups, such as with adult or juvenile harp seals and grey seals.

Meanwhile, CART segregated individuals by pre-defined groups, for example by gender,

even when no significant differences could be established between groups on the chosen

fatty acid.

The differences observed in the patterns issued from the two approaches may be

better understood by examining the characteristics inherent to these methods. CART

examined all fatty acids at once, then proceeds to partition the cases into the defined classes

based on decisions using one fatty acid at a time. By comparison, the PC/Cluster analyses

used all significant information (retained factors) simultaneously in order to reveal

associations (i.e., clusters) among all cases at once, and did so without reference to the

classes involved. This difference in approaches between the two data exploratory methods

was reflected in our study, where the cluster analysis classified adult males and females of

hooded seals and harbour seals as part of the same cluster (i.e., clusters 1 and 9,

respectively), whereas CART identified the specific fatty acids that would achieve

segregation of cases by species, or by gender within each species when directed to classify

these classes.

One benefit of CART over other multivariate analyses is the relaxed restrictions on

the number of entry variables, allowing the inclusion of fatty acids that might be strongly

correlated and whose relevance is not known beforehand. By comparison, PCA usually

benefits from some restriction in the number of variables used, although this is not a strict

111

requirement provided other conditions are met, e.g. with > 100 samples and loadings > 0.6

(Field 2000, Osborne and Costello 2004). Cluster analysis is particularly sensitive to

correlations among variables, and thus can benefit from the use of factors following a PCA

or other means of transforming data and preselecting independent variables. Despite the

flexibility of exploratory methods such as PCA and CART in particular with regards to the

number and correlation among variables, some degree of variable selection is often

warranted. The selection of a reduced set of diet-linked fatty acids has obvious benefits

since some of the fatty acids available to conduct exploratory analyses might be of lesser

value, either because they occur in ‘trace’ (usually < 0.5%) amounts or because they are not

of strong interest for the question being addressed, such as associations with diet sources vs.

environmental or genetic influences.

Initially, some of the classification trees developed using CART were based on fatty

acids found in trace amounts or suspected to be intermediate metabolites, such as 18:2n-7

and 22:5n-3 for the species classifications (Fig. 3.2). The use of 18 instead of 56 fatty acids

as variables resulted in similar classifications (i.e., from 20/143 to 21/143). This pattern

indicated a certain robustness of the CART approach, namely that the patterns revealed

when using the initial database were essentially repeated when using only the subset of fatty

acids. Thus, the similar groupings in both CART models may be due to underlying factors

associated with these fatty acids, rather than to the high variability (and thus potentially

high discriminatory power) inherent to trace fatty acids. These results also brought

confidence that significant patterns of interest in the total database could be revealed during

the variable reduction process of PC/Cluster analysis.

While valuable for exploring among all potential variables, CART can be less useful

for understanding individual cases because these are not identified in the analysis during

112

their placement in the pre-defined classes. PC/Cluster analyses can be better suited for

exploring group associations because they maintain the identity of cases when evaluating

them on the available variables, in this case, their similarity in fatty acid composition. In

addition, PC/Cluster analyses have no requirement to establish classes prior to analysis.

This feature of the PC/Cluster analysis provides it with an advantage over supervised

classifications, such as CART and discriminant function analyses, particularly in the

context of studies whose aim is to reveal unknown and shared classes, as might occur with

dietary overlap among species. When examining this question, the investigator is interested

as much in the similarities as in the differences in fatty acid composition between species or

other classes.

In this study, the PC/Cluster analysis permitted the segregation of a group of harp

seals including adults and sub-adults (cluster 3) and a group of grey seals (cluster 8), also

including both age-classes, from other harp seals (clusters 5 and 6) and grey seals (cluster

5) on the basis of much lower levels of some fatty acids of dietary origin, i.e.,

monounsaturated fatty acids associated with Factor 2 (Table 3.6). In the CART analysis,

the overlap of cases due to their similarities in fatty acid composition was not actually

specified, but rather was implied from those cases that were ‘misclassified’ within the prior-

defined classes such as species and gender (Walton et al. 2000). The ‘failure’ of CART to

correctly classify cases in specific tree models then becomes useful information, as the

‘misclassified’ cases reveal the potential inadequacies of the defined classes. One possible

interpretation is that the misclassifieds in classification treees represent individuals with

similar trophic histories, for example between species, or between genders and age-classes.

In these instances, it may be interesting to evaluate for these hidden (i.e, not pre-defined),

‘functional’ groups.

113

The value of interpreting groupings as suggested from cluster analysis and as

compared to testing for pre-defined classes was revealed through the cross-validations

using discriminant functions analyses. For example, the position of harbour seals and

hooded seals along the discriminant function was inverted when classifications (for both

species and clusters) were performed using PC factors (Figs. 3.6a, 3.9a) or fatty acids (Figs.

3.6b, 3.9b) as classification variables. This reversal revealed the utility of classifying

subgroups such as the harbour seal subadults (seen elsewhere as Cluster 7), apart from the

adult harbour seals whose mean scores were closer to that of the hooded seals (Clusters 9

and 1, respectively) when using the PC factors. Alternatively, this reversal in position may

indicate a greater sensitivity to a loss of information through factor reduction in these two

species. Indeed, the polyunsaturated fatty acids that contributed largely to PC factor 1 were

all positively correlated among each other in the PCA, but some fatty acids were ascribed

differing values (opposing vectors) in discriminant analysis when they were used

separately, i.e., fatty acids 20:4n-3 and the 22:6n-3. A similar situation was observed for

the fatty acids 14:0, 16:0 and 16:1n-7, which all loaded positively on Factor 4, but which

differed in the vector of their contributions in the discriminant functions analysis (Figs. 3.6

and 3.9). These differences in the visualisation of contributing fatty acids vs. PC factors are

of potential consequence in trophic interpretations, as will be discussed later.

The similarity in general patterns detected by CART and PC/Cluster analyses in this

study indicated that the two approaches were useful for the study of fatty acid composition

of marine predators. In some instances, the two techniques provided complementary

information, which emphasised the power of these tools when used in combination. The

strength and weaknesses of the two approaches in their capacity to answer questions of

114

interest such as dietary overlap must be borne in mind while searching for biologically-

meaningful patterns among individuals.

Biological significance of patterns in fatty acid composition

Throughout this investigation, diet-linked fatty acids were consistently chosen as the

primary factor for grouping cases, even when these fatty acids were not necessarily the

most abundant nor the most variable. Among those fatty acids of dietary origin,

polyunsaturated and monounsaturated fatty acids were the most successful variables at

classifying species or age- and gender-classes in CART and constituted the two main

principal components and segregation criteria in the PC/Cluster analysis.

Polyunsaturated fatty acids were used in both CART and the PC/Cluster analysis

(through Factor 1) to broadly discriminate between marine mammal species that reside or

visit the Estuary and Gulf of St. Lawrence, thereby separating resident harbour seals and

belugas from Gulf visitors such as grey, harp, and hooded seals. In general, it might be

proposed that the lower levels of polyunsaturated fatty acids observed in beluga and

harbour seals (especially sub-adults) reflected their history in the estuary while higher

levels in harp and grey seals (especially adults) revealed presumably greater dietary inputs

of marine origin (Ackman 1989, Budge 1999).

This trophic-residency hypothesis is supported by the results from an earlier

investigation, where the same individual marine mammals used in this study were analysed

for their stable carbon (δ13C) and nitrogen (δ15N) isotope ratios (Lesage et al. 2001). The

δ13C values of seasonal visitors to the Estuary and Gulf of St. Lawrence were generally

lower than those of species resident to the Estuary, and were also suggestive of seasonal

movements outside of the Estuary by grey seals and harp seals that were captured within

115

the Estuary. These differences were suspected to reflect the characteristics of carbon

sources of the food webs from the two environments, a larger enrichment in 13C in the

Estuary resulting from a greater productivity and input of particulate matter of terrigenous

origin (Lesage et al. 2001). The large number of prey of that may enter the diet of beluga

and pinnipeds of the St. Lawrence, and the limitations in the number of potential prey that

can be entered in prey source mixing models to calculate diet composition using isotope

ratios (Phillips and Gregg 2001) limited the ability of the authors to evaluate the specific

prey inputs from estuarine vs marine origins, or combinations thereof, using these markers.

Without greater specificity or external validation of inferred diets when using either

stable isotopes or fatty acids, both as individual tracers or combined as a profile, the

trophic-related factors responsible for the general pattern differentiating estuarine residents

and Gulf visitors must await confirmation. Meanwhile, in order to assist with present and

future interpretations, other question of proximate causes might be examined with the

current, limited set of fatty acid data. For example, one of the most obvious findings in this

study was the strong segregation of beluga from the pinniped species on the basis of low

amounts in polyunsaturated fatty acids. While the main hypothesis was to find species-

related, diet-linked differences, we could not discount the uncontrolled factors of sampling

procedure and sample classes, in particular body size, as also affecting the observed fatty

acid compositions.

Samples of beluga blubber were all collected from stranded carcasses and not live

animals. Unlike the present situation with pinnipeds, or even birds and fish, local sampling

and controlled diet studies in captivity have thus far not been possible with belugas. This is

due to logistics, conservation issues, but also biological differences, i.e., blubber

stratification and metabolic specialisation, more-invasive biopsying of thicker dermis or

116

deeper blubber, and the difficulty to control regularity of feeding in belugas. The problems

with sampling stranded animals are not yet well-documented, despite their value as an

alternate and relatively-available source of information, especially for restricted

populations. Documenting comparisons of carcasses for several species of cetaceans and

pinnipeds might be especially valuable as there may be species-specific patterns in the

stratification of fatty acid compositions in blubber tissues (Käkelä et al. 1993, Koopman et

al. 2002, Best et al. 2003b, Olsen and Grahl-Nielsen 2003, Arnould et al. 2005).

Documenting similarities and differences will then assist the application of knowledge

gained from model organisms in captivity (e.g., rats, seals, mink, and fishes) towards

calibrating diet models and will likely result in more accurate as well as more precise

estimations.

Apart from unknown species-based differences, there is also the issue of comparing

samples from studies. The manner of tissue sampling is sometimes suspected to have

unforeseen effects on detected compositions, as in the case of the leaching of fats when

sampling in small amounts, especially when using biopsies (Krahn et al. 2004, Thiemann et

al. 2004b). Leaching may also be of concern when considering the role of poor health or

post-mortem changes on blubber tissues. In these cases, tissue quantity is not the issue so

much as determining the quality and the degree to which further analysis may still be useful

relative to biopsying live or fresher individuals. (Hooker et al. 2001, Evans et al. 2003,

Krahn et al. 2003, Mau 2004). The result is that different pictures may arise during

comparisons. Thus, one study on pinnipeds (harbour seals) suggested relative compositions

went unchanged despite lower absolute levels in emaciated individuals (She et al. 2002).

Similarly, studies of feeding in seabirds and pinnipeds revealed fine-scale, diet-linked

changes in blood composition reflecting the signature of prey species consumed regardless

117

of total lipid content (Cooper et al. 2005, Käkelä et al. 2005). In contrast, a broader study

of blubber compositions in a species of cetacean (harbour porpoise) concluded that diet-

based fatty acid signatures were diminished in emaciated compared to normal individuals,

thereby making it unlikely for longer-term information on diet to be obtained in the poor-

condition individuals (Koopman et al. 2002).

The condition of blubber or other fat stores can be evaluated by measures such as

gross physical signs of tissue decay or an increase in the proportion of free fatty acids in

lipid classes (Ohman 1996, Krahn et al. 2003, Mau 2004). In our study, no general trend in

fatty acid compositions was detected from sampled belugas that were classed as ‘good’

(fresh carcass collected for necropsy) vs. ‘poor’ (tissues collected from individuals in

advanced decomposition), even when examining additional samples that were obtained and

processed more recently (1999-2002, not shown here). One possible explanation for the

similarity in compositions from both ‘good’ and’ poor’ samples is that they were all of the

same type (obtained from carcasses), and thus equally ‘degraded’ relative to what might be

seen in live individuals. Support for this idea may be seen with selected cases. For

example, four male belugas (three adults and one newborn) scored closer to pinnipeds in

the PC/Cluster analysis than they did to all other belugas, mainly because of their higher

abundances in polyunsaturated fatty acids (i.e., Factor 1). Three belugas (not identified)

were also ‘misclassified’ among the seals on this basis in the CART species models. These

individuals may have been in relatively better condition, that is, were sampled more

recently following death, than the other belugas, and possibly more similar to the other

marine mammals sampled in this region or with live-sampled belugas sampled elsewhere

(Dahl et al. 2000). An alternate solution would be to evaluate the fatty acid scores of the

stranded belugas relative to those of normal and stranded cetaceans obtained in one region,

118

for example, by using available data on harbour porpoises (Koopman 2001). In this latter

case, the belugas from this study would score closer to stranded, emaciated porpoises than

to those sampled in normal condition (bycaught in fisheries) or pinnipeds (not shown).

Setting aside the question of sample condition and its effect on the beluga

signatures, it remains of interest that lower amounts of polyunsaturated fatty acids were

observed at once in the largest (hooded seals) and smallest pinniped species (harbour seals)

examined here (Table 3.2). In the case of the harbour seals, this similarity with belugas in

general (i.e., in PCA and CART, but not cluster analysis) suggests that residency in the

estuary is the common factor in their polyunsaturated fatty acid compositions. The closer

proximity of harbour and grey seal sub-adults (cluster 9) further suggests estuarine diet

sources, perhaps as neonates and yearlings did not experience foraging outside of the

estuary relative to adults that may travel farther in the winter months (Lesage 2004).

Interestingly, the relatively low amounts of ‘marine-linked’ fatty acids such as

20:5n-3 and 22:6n-3 in the resident marine mammals from the estuary was not necessarily

counter-balanced by higher amounts of more ‘estuarine-linked’ n-6 series fatty acids. This

could be taken to suggest that resident prey species did not make significant contributions

of lipids to the diet of belugas or sub-adult seals. For example, fatty acids 18:3n-3 and

18:2n-6 may be derived from terrigenous inputs (plants and detritus), and are expected in

greater amounts in freshwater prey including American eels (Anguilla rostrata), salmonids,

or common estuarine invertebrates such as sand shrimps (Crangon septemspinosa)

(Ackman and Hooper 1973, Smith 1999, McKenzie 2000).

Eels are a special case, as they represent a large and fatty prey item (averaging 14%

lipid by wet weight, see Chapter 2) and even their occasional consumption would represent

a significant source of lipid-soluble pollutants such as Mirex and other organochlorine

119

pesticides that are detected in substantial concentrations in the St. Lawrence beluga (Hickie

et al. 2000, Falandysz et al. 2002). However, detecting evidence for eels in diet is greatly

diminished by historical and seasonal factors. Firstly, the eel population of the St.

Lawrence watershed has undergone a drastic decline in recent decades, thereby likely

reducing their potential as prey (Bérubé and Lambert 1999, Wirth and Bernatchez 2003).

More importantly, eels spend their adult lives in streams (hence the ‘estuarine-linked’ diet

and contamination) and are only present in the Estuary during a short period in October, at

which time they undertake their final spawning migration to the ocean. While reports are

rare, belugas have been observed feeding on eels in the Saguenay River during this brief

window of opportunity. However, as the samples from belugas and seals presented here

were obtained in spring and summer, it is not expected that these individuals would show

any evidence of eel consumption dating from the preceding autumn.

Normally, fatty acid signatures in blubber as related to diet are assumed to act as

‘medium-term’ tracers that remain in evidence for some weeks and months after

assimilation (Kirsch et al. 2000, Iverson et al. 2004). More-recalcitrant tracers may be

necessary in order to reveal the diet history of past seasons and years in a sampled predator.

Synthetic contaminants are one example, as certain organisms (e.g., mammals) may have a

limited capability to modify or eliminate such compounds. In a related study, samples from

the same series of prey and predators were examined for organochlorine contaminants. The

results suggested that the ‘estuarine’ (stream- and shore-based) eels, smelt (Osmerus

mordax), and smooth flounder (Pleuronectes putnami) were associated with a separate food

web from that shared by other fishes or harbour seals and beluga whales residing in the St.

Lawrence Estuary (Law et al. 2003). Such a finding, if confirmed with other tracers, might

120

account for the relatively low amounts of ‘estuarine-linked’ fatty acids in belugas and

harbour seals.

After the polyunsaturates, the monounsaturated fatty acids were selected as the

next-most important variables by CART and the PC/Cluster analysis (i.e., Factor 2) to

discriminate between classes including species, age, and gender. Specifically, 18- and 20-

series mononunsaturated fatty acids were found in higher amounts (and conversely, lower

amount of polyunsaturated fatty acids) in the larger species such as the beluga (22:1n-11)

and hooded seal (18:1n-9) than in the smaller species. Similarly, diet-linked

monounsaturates (Iverson et al. 2004) were more abundant in adults relative to juveniles, as

in the cases of harp seals (22:1n-9) and grey seals (20:1n-7).

However, the effects of diet or physiological processes on the abundance of

monounsaturated fatty acids are usually more difficult to isolate as they are both abundant

and amenable to modification in adipose tissues in comparison to many polyunsaturates

(Lin and Conner 1990). Harbour seals were perhaps the clearest example of this difficulty,

as the optimal fatty acid selected in CART for segregating adults from subadults was the

common and abundant fatty acid 16:1n-7 (Bremer and Noren 1982, Ackman 1989, Iverson

1993, Leonard et al. 2003, Soriguer et al. 2003). Unfortunately, only harbour seals were all

sampled from the same period and region for all age-classes of their species. Harp seal sub-

adults were sampled in the Estuary, whereas all but three adults came from the Gulf. For

grey seals, yearlings were sampled in the estuary over the summer while adults were

obtained from winter breeding grounds. However, across all three pinniped species, a

common trend appears to be the reduced mean amounts of 20-series mononunsaturates,

most often for 20:1n-9 and 22:1n-11, in sub-adults compared to adults, even when this was

not highlighted from the multivariate analyses, as in the case of harbour seals here (not

121

shown). In other studies, when comparing individuals within a specific region, similar

differences for the two age-classes using 20-series fatty acid compositions were evident in a

large-scale study of grey seals, but were not seen in smaller studies of harbour seals or harp

seals (Andersen et al. 2004, Falk-Petersen et al. 2004, Beck et al. 2005).

One possibility for the observed differences in this and other study is that larger

individuals may have relatively greater deposits of fat compared with smaller individuals.

Such a trend has been suggested to occur in large fish such as salmon, herring, turbot, and

cod, as these species are often fattier and enriched in either polyunsaturated or

monounsaturated fatty acids relative to smaller individuals of their species (Iverson et al.

2002, Veefkind 2003).

Along with the influence of physiology and related to this trend in allometry is the

potential influence on fatty acid patterns from foraging behaviour associated with dive

profiles (Schreer et al. 2001, Beck et al. 2005). Behaviour and size may play a role in fatty

acid compositions if larger individuals forage differently, consuming prey of lesser quality

or greater size and species diversity relative to smaller predators (Beck et al. 2003). By

consuming larger fish specimens which have themselves accumulated more of certain fatty

acids, as in the examples described above of herring with 22:1n-11 and cod with 22:6n-3,

large predators such as adult grey seals will then also reflect this accumulation with their

body size relative to smaller marine mammals. Deep-diving by larger predators may also

provide increased access to monounsaturated-rich (i.e., 20:1n-9, 22:1n-11) prey including

squid, redfish, and Greenland halibut (see Chapter 2), which might account for the elevated

levels of these fatty acids detected in hooded seals.

In addition to biology, behaviour then might be involved with the recurring trends of

small, but at times significant, differences in diet-linked fatty acids (polyunsaturates and

122

monounsaturates) in sexually-dimorphic species (Beck et al. 2005). Thus, size-linked

diving and foraging behaviour might then be involved in the differences seen in this study

between the males and females of beluga whales, hooded and grey seals, and to a lesser

extent harbour and harp seals. Alternately, gender-based differences in metabolism rather

than diet might be involved. Unfortunately, there is a tendency in biology to assume that

gender has no role, or that the male organism can serve as unspoken reference for the

physiology of a given species. A recent diet study on human nutrition revealed similar

patterns as seen in this study, with female metabolic patterns resulting in slightly higher

levels of certain minor fatty acids (e.g., 14:0, 18:2n-6) than males (Burdge and Calder

2005). Separating foraging history from physiological allometric- and gender-based

influences will continue to present challenges towards generalising diet-linked signatures of

fatty acids, but also opportunities to further discriminate among identified classes

(Andersen et al. 2004, Beck et al. 2005).

Finally, apart from the trends explored here for individuals of different classes and

species, there is also a need to consider the type of samples used, and the way fatty acid

compositions are investigated in general, in order to draw the most information from such

studies. Firstly, the blubber samples analysed in this study were almost exclusively

composed of neutral lipids (triacylglycerols) which are comprised of fatty acids as

assimilated from the total lipids of ingested prey (Ackman 1989, Christie 2003). Secondly,

the use of blubber tissue highlighted dietary differences through the multivariate treatments

used in this and other fatty acid signature studies (Walton et al. 2000, Budge et al. 2002,

Iverson et al. 2002). Comparisons using a set of diet-linked fatty acids distinguishes these

results from studies examining fatty acids which focused on the success of categorising

individuals, even when this was performed on the basis of lesser-known fatty acids in trace

123

amounts, such as the initial CART model (see also Iverson et al. 1997b, Møller et al. 2003,

Krahn et al. 2004).

As an example of the pitfalls with interpreting only trace-level fatty acids, we may

look at the polyunsaturate 20:4n-6, a fatty acid which is useful as a diet tracer at the

planktonic level (Dunstan et al. 1993). In mammals, however, 20:4n-6 is principally

associated with metabolic functions and is conserved in phospholipids of cell membranes,

including those of adipose tissues (Durnford and Shahidi 2002, Lapillonne et al. 2002, Best

et al. 2003a, Rapoport 2003, Marangoni et al. 2004). The result is that 20:4n-6 is often

only detected in very small, stable amounts relative to other diet-linked fatty acids in the

blubber, and was subsequently not considered in the subset of fatty acids for statistical

analyses. Similarly, while some studies have successfully used monousaturated fatty acids

(e.g., copepod-derived 20:1n-9) to infer specific plankton contributions in lower-level

predators (Falk-Petersen et al. 2000, Dahl et al. 2003), their abundance in mammalian fats

is not necessarily a reflection of the importance of plankton as direct prey inputs to tertiary-

level predators (Dahl et al. 2000, Andersen et al. 2004).

While our understanding of lipid biology and digestive assimilation from observing

free-living individuals is still being developed, ecological inferences are possible because of

the insights obtained from empirical investigations (Kirsch et al. 1998, Kirsch et al. 2000),

along with feeding observations (Bowen et al. 2002), and the results of diet modeling (i.e.,

QFASA, Iverson et al. 2004). Controlled feeding trials using captive animals, and an

increasing number of studies on species in the wild, have demonstrated the value of fatty

acid profiles in discriminating regional dietary patterns among populations of marine

mammals, including those of fishes, molluscs and other invertebrates (Skerratt et al. 1995,

Cripps and Atkinson 2000, Freites et al. 2002, Lea et al. 2002a, Bradshaw et al. 2003,

124

Walton and Pomeroy 2003). Identifying differences with which to distinguish populations

is a reliable function of analysing fatty acids, among other biomarkers (Westgate and Tolley

1999, Møller et al. 2003, Olsen and Grahl-Nielsen 2003). In some cases, this relative ease

with which patterns of differences are seen among samples may in fact become a hindrance.

This is because success at detecting differences may make it easy to overlook the

difficulties at elucidating the influences such as diet, and the necessity to compensate for

factors including the assimilation of prey fatty acids in predator tissues (Kirsch et al. 1998,

Iverson et al. 2004, Thiemann et al. 2004a).

Summary of potential diets

While this survey of the patterns and influences on fatty acid compositions is

neither exhaustive nor conclusive, the results can be summarised given what is generally

believed regarding the diet of these marine mammals, principally from existing knowledge

on foraging behaviour and digestive tract contents. For example, the classifications

presented here are not inconsistent with belugas foraging predominantly on small marine

pelagics, linked to the copepod food web on the basis of their amounts in 20-series

monounsaturated fatty acids. These pelagics may include sand lance (Ammodytes sp.) and

capelin (Mallotus villosus), which are species that have been previously identified as among

the dominant contributions to foraging in sampled belugas from the St. Lawrence during the

1930s (Vladykov 1946).

Harbour, harp, and grey seals are another group that are consumers of small forage

species, however their relatively higher concentrations of marine and demersal species-

linked fatty acids (including 20:1n-9, 20:5n-3 and 18.1n-7) relative to the belugas might

point towards to their consumption of additional species, or perhaps larger specimens, as

125

was further witnessed by higher amounts of these fatty acids in the clusters of adults

relative to juvenile harbour seals. Marine pelagic and demersal fishes such as herring, cod,

and flatfishes are reported to be among the principal prey shared by these pinnipeds (Bowen

and Harrison 1994, Hammill and Stenson 2000). Support for the suggested prey may also

be seen from stomach contents which had samples dominated by crustaceans and capelin in

harp seals, while other samples from grey seals had mostly herring, cod, turbot, and plaice

(Hammill et al, unpubl.).

Unlike the considerable overlap in fatty acid compositions, and presumed diet,

between several groups of pinnipeds, hooded seals were notably classified apart from the

other seals. However, hooded seals may also represent a special case, as their biology and

behaviour is arguably the least-understood of all the predators examined in this study

(Hammill et al. 1995, Folkow and Blix 1999, Schreer et al. 2001). Nonetheless, using

strictly a diet-based approach, it can be postulated that the low levels of polyunsaturated

and elevated levels of monousaturated fatty acids observed for this large-bodied predator

were consistent with greater inputs of deepwater rather than coastal or epipelagic prey,

relative to the other pinnipeds sampled here. Prey species available offshore and in deeper

waters include redfish (Sebastes sp.), boreal squid (Gonatus fabricii), and Greenland halibut

(Reinhardtius hippoglossoides), as has been previously reported in preliminary analyses of

some stomach content samples of hooded seals (Hammill et al. 1995, Hauksson and

Bogason 1997, Kapel 2000, Potelov et al. 2000).

126

3.5. Conclusion

In this study, the examination of similarities and differences in the fatty acid

composition of blubber sampled from beluga and four species of pinnipeds from the

Estuary and Gulf of St. Lawrence resulted in several groupings of individuals suggesting

the influence of dietary factors. The parallel use of several analytical approaches helped in

the interpretation of the results, as they variously highlighted consistencies, but also

inconsistencies, in the patterns observed in fatty acid compositions for these diverse groups.

These findings emphasize the benefit of using a combination of approaches not only in the

exploration of the data, but also in the validation of the classification results, especially in

the absence of direct confirmation of diets from observations and stomach contents. The

comparative approach is particularly important in multivariate analyses for which several

options exist involving the use of cases and variables, but also for viewing output, whose

suitability in revealing patterns may vary between studies. It also suggested, however, that

careful data analysis was by itself insufficient to provide a detailed understanding of

predator fatty acid profiles as they related to the consumption of prey species. As has often

been the case with related groups that are believed to share diets, specific dietary inferences

will first require more analyses of the differing predator classes, such as gender and size.

Also of direct benefit would be additional information on the factors involved, including the

condition of tissue samples, the metabolism of different classes of predators, and the

evaluation of the variability in prey species signatures from the region under investigation.

Ultimately, ‘calibration’ from such complementary investigations using fatty acid

signatures as trophic biomarkers will lead to increased knowledge of the different and

shared dietary sources among marine mammals. This knowledge will be especially

127

valuable with regards to understanding trophic interactions between the more-abundant

harp and grey seals as compared with at-risk populations in the case of the belugas of the

St. Lawrence Estuary.

128

Table 3.1. Marine mammals sampled in the Estuary and Gulf of St. Lawrence between

1996–1998.

Species Region Age class Total Male Female

Phoca vitulina Estuary Adult 11 5 6

Harbour seal Sub-adult 24 12 12

Newborn 8 3 5

Pagophilus groenlandicus Gulf Adult 15 7 8

Harp seal Estuary Adult 3 0 3

Sub-adult 13 6 7

Halichoerus grypus Gulf Adult 16 8 8

Grey seal Estuary Sub-adult 15 7 8

Cystophora cristata Gulf Adult 19 10 9

Hooded seal

Delphinapterus leucas Estuary Adult 17 9 8

Beluga whale Sub-adult 1 a a

Newborn 1 1 0 aGender unknown

129

Table 3.2. Fatty acid compositions of five marine mammal species sampled in the

Estuary and Gulf of St. Lawrence. Values are means by mass % (± SEM) for

33 fatty acids with overall mean abundances ≥ 0.2%. The 18 fatty acids

selected for use in statistical analyses are indicated in bold.

Hooded seal

(n = 19) Beluga

(n = 19 ) Grey seal (n = 31)

Harp seal (n = 31)

Harbour seal (n = 43)

Groups Fatty acid Mean SEM Mean SEM Mean SEM Mean SEM Mean SEM Saturated C5:0 0 0 0.87 1.83 0 0 0 0 0 0 fatty acids C12:0 0.06 0.02 0.72 0.61 0.11 0.02 0.1 0.04 0.14 0.03 (SFA) C14:0 3.58 0.59 6.76 1.52 4.43 0.51 5.42 0.74 4.02 0.48 C15:0 0.2 0.02 0.3 0.05 0.26 0.03 0.25 0.04 0.23 0.02 ISO15:0 0.12 0.01 0.34 0.12 0.15 0.02 0.18 0.04 0.12 0.02 C7ME16:0 0.23 0.03 0.21 0.04 0.28 0.05 0.32 0.05 0.25 0.03 C16:0 6.49 0.85 8.73 1.84 8.06 1.28 7.79 1.6 8.84 1.46 C18:0 1.07 0.15 1.53 0.57 0.9 0.16 0.95 0.22 0.86 0.13 Mono- C14:1n-9 0.14 0.01 0.44 0.43 0.14 0.03 0.19 0.04 0.15 0.03 unsaturated C14:1n-7 0.04 0.01 0.36 0.32 0.07 0.01 0.05 0.02 0.09 0.02 fatty acids C14:1n-5 0.68 0.07 0.93 0.89 1.04 0.23 0.71 0.33 1.53 0.45 (MUFA) C16:1n-11 0.56 0.09 0.61 0.68 0.46 0.08 0.44 0.1 0.6 0.11 C16:1n-9 0.41 0.04 1.24 0.68 0.4 0.09 0.25 0.07 0.58 0.1 C16:1n-7 12.55 1.26 12.69 4.86 13.94 1.61 14.85 3.19 18.93 3.66 C18:1n-11 4.09 1.1 3.88 1.33 4.08 0.97 2.58 1 4.03 1.53 C18:1n-9 22.48 3.15 17.64 2.5 15.02 3.1 10.48 2.33 17.5 1.49 C18:1n-7 4.75 0.47 3.54 0.59 4.71 0.9 3.57 1.01 4.47 0.31 C18:1n-5 0.45 0.03 0.32 0.06 0.49 0.07 0.48 0.09 0.4 0.03 C20:1n-11 2.37 0.41 4.29 1.12 2.18 0.58 1.62 0.62 2.16 0.64 C20:1n-7 0.95 0.17 0.61 0.22 0.86 0.26 0.75 0.33 0.49 0.09 C20:1n-9 13.56 1.45 8.79 2.2 8.77 2.29 11 3.83 7.94 1.79 C22:1n-11 5.09 1.13 8.79 3.46 3.32 1.84 5.69 3.12 2.97 1.23 C22:1n-9 1.09 0.26 0.95 0.43 0.65 0.28 0.89 0.5 0.41 0.12 Poly- C16:3n-4 0.26 0.04 0.21 0.06 0.26 0.06 0.19 0.05 0.27 0.06 unsaturated C16:3n-6 0.43 0.1 0.6 0.13 0.55 0.14 0.71 0.12 0.56 0.11 fatty acids (PUFA) C18:2n-6 1.11 0.16 0.86 0.13 1.24 0.16 1.19 0.29 0.85 0.09 C18:3n-3 0.34 0.07 0.29 0.1 0.46 0.07 0.49 0.13 0.32 0.09 C18:4n-3 0.71 0.17 0.27 0.09 1.09 0.27 1.86 0.87 0.66 0.13 C20:4n-3 0.47 0.07 0.28 0.07 0.54 0.08 0.53 0.14 0.28 0.06 C20:5n-3 4.15 0.78 1.96 0.67 5.88 0.88 8.73 2.06 4.89 1.13 C20:4n-6 0.25 0.07 0.25 0.06 0.39 0.14 0.29 0.08 0.47 0.15 C22:5n-3 2.3 0.33 2.59 1.19 5.23 0.57 4.43 0.93 4.22 0.76 C22:6n-3 5.96 0.72 4.75 1.53 9.77 1.43 8.91 1.78 7.06 1.21

130

Table 3.3. Significant loadings (>0.5) of 17 fatty acid logratio variables on the four

principal components (PC) that were retained (Eigenvalues > 1). The

maximum loading for each fatty acid variable is highlighted in bold. Also

shown is the % of variance (Varimax-rotated) accounted for by each PC.

Kaiser-Meyer-Olkin measure of sampling adequacy was 0.772.

Principal component Normalized logratio 1 2 3 4

18:4n-3 0.938 20:5n-3 0.931 20:4n-3 0.916 22:6n-3 0.878 18:2n-6 0.834 18:3n-3 0.829 20:1n-7 0.582 22:1n-9 0.921 22:1n11 0.903 20:1n-9 0.786 18:1n-9 0.907 18:1n-11 0.775 18:1n-7 0.559 0.726 20:1n-11 0.587 0.653

14:0 0.831 16:0 0.693

16:1n-7 0.548 0.560 % variance explained 36.84 19.91 19.04 12.18

% cumulative variance 36.84 56.75 75.79 87.97

131

Table 3.4. Results of one-way ANOVAs (F4, 138, P < 0.005) for differences among

species groups for each PC factor. Values in the same column not sharing a

common superscript letter are significantly different (post hoc Tukey HSD

multiple comparison of means, overall a = 0.05). Species groups are:

hooded seals (Cc), beluga whales (Dl), harp seals (Pg), grey seals (Hg), and

harbour seals (Pv).

Factor Species group N 1 2 3 4

Pg 31 0.87a 0.32a -1.12c 0.35a Hg 31 0.71a -0.04ab 0.27a -0.28b Cc 19 -0.08b 0.46a 0.42a -1.44c Pv 43 -0.23b -0.50b 0.63a 0.45a Dl 18 -2.03c 0.25a -0.45b 0.33ab

132

Table 3.5. Results of ANOVA (F8, 134, P < 0.001) to test differences among clusters on

each PC factor. Mean values in the same column not sharing a common

superscript letter are significantly different (post hoc Tukey HSD multiple

comparison of means, overall a = 0.05). The number of members for each

cluster is indicated following each species: hooded seals (Cc), beluga whales

(Dl), harp seals (Pg), grey seals (Hg), and harbour seals (Pv).

Factor Cluster Members (n) 1 2 3 4

1 Cc (17) -0.14bc 0.37ab 0.46b -1.54c

2 Dl (10) -1.73d 0.35ab 0.22b 1.43a 3 Pg (6), Hg(1) 1.10a -2.06d -1.43c 0.08b 4 Dl (9) -2.29d 0.08b -1.20c -0.90c

5 Hg (17), Pg (8), Cc (2) 0.83a 0.73ab 0.14b 0.18b 6 Pg (17) 0.64a 0.96a -1.59c 0.36b 7 Pv (29), Hg (3) -0.30c -0.81c 0.36b 0.53b 8 Hg (8), Pv (1) 0.80a -1.19cd 0.55ab -1.15c

9 Pv (13), Hg (2) 0.13b 0.29b 1.19a 0.13b

133

Table 3.6. Means (±SEM) for the subset of 18 fatty acids (normalized over 100% for each individual) on the nine marine mammal

cluster groups resulting from the hierarchical cluster analysis on 4 PC factors derived from the 17 logratios.

Cluster 1 2 3 4 5 6 7 8 9

n 17 10 7 9 27 17 32 9 15 Mean SEM Mean SEM Mean SEM Mean SEM Mean SEM Mean SEM Mean SEM Mean SEM Mean SEM

14:0 3.84 0.15 8.82 0.51 6.44 0.44 6.71 0.45 5.48 0.10 6.27 0.19 4.91 0.08 4.55 0.20 4.23 0.11 16:0 7.21 0.23 9.48 0.70 11.35 0.86 10.67 0.44 8.40 0.30 9.03 0.28 10.96 0.21 9.34 0.45 8.28 0.22

16:1n-7 13.55 0.27 19.01 1.92 20.98 1.74 10.28 0.85 16.31 0.37 15.33 0.63 23.16 0.68 16.49 0.72 17.64 0.51 18:0 1.21 0.04 1.27 0.12 1.21 0.10 2.27 0.09 0.94 0.03 1.16 0.05 1.06 0.02 1.12 0.08 0.87 0.03

18:1n-11 4.53 0.31 5.07 0.43 1.47 0.20 3.86 0.55 4.88 0.18 2.88 0.11 3.96 0.21 3.68 0.38 6.45 0.29 18:1n-9 25.48 0.67 21.55 1.33 13.84 1.17 19.26 0.72 15.27 0.33 10.10 0.31 19.69 0.31 22.39 0.81 20.55 0.41 18:1n-7 5.34 0.10 3.78 0.18 6.24 0.32 4.41 0.15 4.55 0.10 3.34 0.16 5.26 0.08 6.55 0.22 5.11 0.11 18:2n-6 1.22 0.05 1.00 0.03 1.69 0.18 0.98 0.07 1.40 0.03 1.22 0.05 1.00 0.03 1.52 0.08 1.06 0.03 18:3n-3 0.38 0.02 0.36 0.03 0.58 0.05 0.30 0.05 0.55 0.02 0.53 0.05 0.36 0.02 0.55 0.03 0.43 0.03 18:4n-3 0.73 0.04 0.27 0.03 2.07 0.29 0.34 0.04 1.49 0.07 2.28 0.30 0.79 0.04 1.04 0.07 0.81 0.03 20:1n-11 2.66 0.11 4.27 0.33 0.83 0.09 5.66 0.29 2.66 0.13 1.98 0.08 2.17 0.09 2.08 0.12 3.21 0.15 20:1n-9 14.70 0.35 8.88 0.77 6.09 1.03 11.40 0.37 12.44 0.45 14.79 0.80 8.44 0.28 7.74 0.33 10.90 0.37 20:1n-7 1.04 0.05 0.54 0.05 0.70 0.09 0.88 0.05 0.94 0.05 0.95 0.11 0.54 0.02 1.12 0.13 0.71 0.03 20:4n-3 0.51 0.02 0.29 0.03 0.55 0.03 0.35 0.02 0.61 0.02 0.64 0.05 0.34 0.02 0.64 0.05 0.38 0.02 20:5n-3 4.43 0.19 2.34 0.22 12.93 1.18 2.19 0.28 7.56 0.29 9.09 0.35 5.52 0.24 6.69 0.44 6.10 0.25 22:1n-11 5.42 0.27 7.95 1.23 2.26 0.70 12.34 0.65 5.11 0.30 9.09 0.51 3.14 0.22 1.34 0.15 4.06 0.38 22:1n-9 1.18 0.07 0.77 0.11 0.45 0.12 1.43 0.12 0.91 0.06 1.35 0.12 0.44 0.02 0.53 0.08 0.57 0.03 22:6n-3 6.59 0.20 4.36 0.43 10.32 0.58 6.65 0.33 10.49 0.35 9.98 0.64 8.26 0.28 12.63 0.51 8.62 0.46

134

Figure 3.1. Approximate sampling locations in the Estuary and Gulf of St. Lawrence.

Beluga carcasses were recovered along the shores of the lower Estuary

(dashed outline). Seals were from (1) Bic, for harbour and grey seals

(Estuary), (2) Godbout, for harp seals (Estuary), (3) near the Magadalen

Islands for breeding harp seals (Gulf), (4) adjacent to Prince Edward Island

(P.E.I.) for breeding hooded seals (Gulf), and (5) near Port Hood, Nova

Scotia for breeding grey seals (Gulf).

135

Figure 3.2. Classification trees for five different species of marine mammals (n = 143)

collected in the Estuary and Gulf of St. Lawrence grouped by species: (a)

initial classification using all available (56) fatty acids, (b) classification

produced while using the reduced set of 18 fatty acids. Fatty acids in

rounded boxes represent variables and the level (% amount) chosen to split

cases. The number of individuals of each species are listed in the terminal

136

nodes (squared boxes), with the numerically dominant species in bold and

the ‘misclassified’ individuals in normal type. Species classes are beluga

(Dl), harbour seals (Pv), harp seals (Pg), grey seals (Hg), and hooded seals

(Cc). Total misclassification rate was 10 and 11 of 143 individuals for (a)

and (b) respectively.

137

Figure 3.3. Classification trees by gender for adult male (M) and female (F) beluga (a),

hooded seals (b), harp seals (c), grey seals (d) and harbour seals (e and f) as

determined by CART while using the set of 18 fatty acids.

138

Figure 3.4. Classification trees by age-class (P = neonates; J = sub-adults; A = adults) for

(a) harp seals, (b) grey seals, and (c) harbour seals, as determined by CART

while using the set of 18 fatty acids.

Figure 3.5. Varimax-rotated PC factor scores of the five marine mammal species on a

matrix of of 4 PCs obtained using 17 fatty acid logratio variables, with

species groups as marked. A separate label was added for a newborn beluga

(black cross) and three beluga males (blue halos) that scored away from

other belugas and towards the pinnipeds on PC 1.

140

a)

b)

Figure 3.6. Discriminant functions analysis of the five species of marine mammals using

(a) scores on the four factors retained from the principal components (PC)

analysis, and (b) the 17 fatty acid logratios as input variables. Group

centroids (star) and 95% confidence intervals are indicated for each species.

141

Variables (PCs or fatty acids) are projected (rescaled to 200% for visibility)

according to their relative contributions on the two discriminant functions.

The cross-validated classification success rate was 84% and 99% (120/143

and 142/143 individuals correctly classified) for (a) and (b), respectively.

Figure 3.7. Dendrogram of hierarchical clustering with Ward’s method using the four

principal component factors as input variables. Horizontal scale denotes the

increase in within-cluster’s variance, rescaled from 0 to 25. Vertical dashed

143

line highlights the joining of cases into a nine-cluster solution. Individual

members are summarised in each cluster by their species: hooded seals (Cc),

beluga (Dl), grey seals (Hg), harp seals (Pg), and harbour seals (Pv). Hooded

seals comprised only adult individuals. Belugas were all adults except for

one individual each in clusters 2 and 4. Beluga and hooded seals are sub-

labelled by gender: female (F), and male (M). Other pinnipeds are sub-

labelled by age-class: adult (A), sub-adults (J), and neonates or pups (P). The

highest aggregation into two clusters is suggested by the labelled boxes for a

grouping of belugas and principally sub-adults seals from the Estuary relative

to all other pinnipeds from the Estuary and the Gulf.

144

Figure 3.8. Mean diamond scatterplots of scores for all individuals by cluster number on

PC factors 1 through 4 (a-d). Diamonds indicate cluster group means ± 95%

confidence intervals (vertically) in proportion to group size (horizontally).

Scores are marked by species: beluga (open circles), hooded seals (triangles),

harbour seals (open boxes), grey seals (x), and harp seals (crosses).

145

a)

b)

Figure 3.9. Discriminant function analysis of the classification results for the nine

clusters using (a) the four principal components (PC) or (b) the set of 17 fatty

acid logratios as input variables. Large circles denote the 95% confidence

interval centroids of mean scores (stars) for clusters 1 through 9. Variables

146

are projected (rescaled to 200% for visibility) according to their relative

contributions on the two discriminant functions. The overall classification

success rate was (a) 97% (139/143 individuals correctly classified) and (b)

93% (133/143 correct assignments).

147

Chapitre 4

Conclusions générales

148

4.1. Sommaire des résultats

Comparaisons de proies potentielles

L’étude du régime alimentaire des espèces à l’aide des signatures d’acides gras est

un objectif intéressant mais qui comporte plusieurs étapes dont les conclusions sont

sensibles à l’interprétation. Une telle compréhension approfondie des profils d’acides gras

de l’ensemble des proies potentielles des prédateurs bénéficiera de l’identification des

signatures d’acides gras alimentaires des prédateurs. Dans une étape importante vers la

construction de modèles de régime alimentaire, plus de 60 espèces de proies potentielles

des mammifères marins (baleine blanche, ou béluga), ont été analysées pour leur teneur en

lipides et leur composition en acides gras.

En plus d’identifier les profils d’acides gras caractéristiques des espèces, la

variabilité intraspécifique a également été étudiée. Plusieurs espèces-clés, particulièrement

les espèces pélagiques, ont démontré de fortes variations en termes de contenu et de qualité

des acides gras selon la taille dans le cas du hareng et du poulamon, et selon la saison ou le

lieu d’échantillonnage dans le cas de l’éperlan et du capelan. Pour ces espèces, un

échantillonnage répété est donc indiqué afin de capter ces fortes variations qui se cachent

dans les valeurs moyennes par espèce. D’autre part, plusieurs espèces de fond, telles que

les crevettes et les poissons plats, démontraient une variabilité beaucoup moindre. Ainsi, un

échantillonnage exhaustif chez ces espèces n’est pas nécessairement utile afin d’améliorer

les données déjà compilées. Néanmoins, les résultats présentés ici restent provisoires

compte tenu des difficultés d’échantillonnage et de l’absence d’expériences de

manipulation alimentaire.

149

Comparaisons de mammifères marins

En se basant sur les profils d’acides gras, nous avons réussi une classification

efficace des prédateurs en haut de la chaîne trophique (quatre espèces de pinnipèdes et une

de cétacé). Ces prédateurs partagent au moins en partie leurs aires de distribution et

d’alimentation dans les écosystèmes marins du Saint-Laurent. Certains acides gras

d’origine alimentaire se sont révélés des facteurs importants pour faciliter la discrimination

des espèces, et même des classes de taille, de sexe ou d’âge. Les différences d’alimentation

des individus seraient donc un facteur important dans l’établissement de ces patrons. La

comparaison d’un nombre plus élevé d’individus des différentes classes pourrait aider à

établir l’importance relative de l’influence du régime alimentaire, de la génétique et de

l’environnement dans les profils d’acides gras observés.

En plus de confirmer la pertinence de certains acides gras comme indicateurs de

groupes trophiques de ces mammifères marins, cette étude a également mis en évidence

certains avantages mais aussi des défis dans l’exploration des données. Ainsi, l’application

judicieuse d’une série de traitements mathématiques (transformations en ratios

logarithmiques des acides gras généralement les plus abondants) et de méthodes

exploratoires des données (arbres de classification, analyses en composantes principales et

de groupements, validées par des analyses discriminantes) a permis d’obtenir des résultats

robustes. L’obtention de résultats semblables suite à l’application des diverses méthodes

était encourageante, surtout considérant la diversité des résultats possibles lors de

150

l’exploration des signatures d’acides gras des mammifères marins, basée sur un si grand

nombre de variables.

4.2. Approches futures

Il existe actuellement un grand intérêt vers l’utilisation des biomarqueurs pour

apporter des informations sur la diète selon plusieurs échelles et scénarios, où l’information

s’avère difficile à obtenir autrement, comme c’est le cas chez le béluga du Saint-Laurent.

Avec cet intérêt, il demeure utile de dénombrer les voies qui aideront à augmenter

l’information vers l’objectif ultime d’identifier plus définitivement les patrons alimentaires

parmi les espèces de mammifères marins dans le temps et l’espace. En continuant le thème

des travaux présentés ici et par d’autres recherches dans le domaine des lipides marins, on

pourrait proposer trois axes de recherche promettant de répondre aux lacunes afin de nous

acheminer vers cet objectif. En somme, ces axes sont 1) la continuation d’une approche à

plusieurs facettes utilisant des biomarqueurs d’acides gras, 2) le développement de

techniques analytiques complémentaires en traceurs trophiques et 3) la validation

généralisée des modèles tels que QFASA. Aucune de ces approches ne guarantit

l’obtention des solutions désirées, mais leur application soignée demeure notre meilleur

espoir pour raffiner notre compréhension, et est en même temps probablement plus efficace

que de suivre le mantra (ou la plainte), peut-être trop commun en écologie, de faire appel à

des études avec ‘encore plus d’échantillons’.

L’approche à plusieurs facettes

Cette approche est presque une philosophie : logique à recommander et à considérer

importante même si ce n’est pas toujours évident de la pratiquer (Iverson et al. 2004,

151

Thiemann et al. 2004a). Pour paraphraser Iverson et al. (2004), on pourrait la résumer ici

comme l’acquisition de l’information écologique par les outils vidéo ou le repérage par

satellite (Bowen et al. 2002), l’emploi des techniques chimiques pour les analyses et les

tissus appropriés (e.g., Budge et Iverson 2003, Cooper et al. 2005) et l’analyse statistique

robuste, pour balancer les explorations de données avec les confirmations. Ce chemin

servira à mieux cibler les informations utiles à d’éventuels tests d’hypothèses. Les raisons

pour lesquelles cette approche n’est pas plus répandue jusqu’à présent sont liées aux coûts

et au temps nécessaire pour entreprendre une étude ayant plusieurs phases d’analyse, mais

aussi à la difficulté qu’il y a à se familiariser avec les multiples domaines qui touchent un

tel sujet ‘écosystémique’. Pourtant, toutes les études ne pourraient pas être de grands

survols (i.e., Budge et al. 2002, Iverson et al. 2002), et il demeurera que des avancées

seront possibles avec la continuation de petites études examinant des aspects spécifiques du

métabolisme et de l’écologie lipidique, si elles sont menées dans l’esprit de garder en tête le

contexte de l’organisme et son environnement.

Les traceurs complémentaires à venir

Comme c’était suggéré dans l’introduction générale, il y a une histoire et de l’intérêt

à développer des méthodes qui se serviront de traceurs complémentaires (Volkman et al.

1998, Ackman 2002, Christie 2003). Il y a lieu de mettre au point de nouveaux éléments de

retraçage, comme via les isotopes stables et les radioactifs moins connus. Une autre

possibilité est de combiner des isotopes communs, tels que ceux du carbone, de l’azote et

du soufre, avec l’intention de produire une ‘signature’ élémentale comme on a vu se faire

avec les acides gras. Leur potentiel demeurera le produit des raffinements de la technologie

analytique, lesquels permettront de souligner de nouvelles sources pour les différences de

152

rapports dans les éléments individuels, ou dans une suite, retrouvés dans un environnement

précis ou dans un tissu spécifique d’organisme.

La validation des modèles : un leçon contemporaine

Après avoir écarté ci-dessus le potentiel de recherche de l’information reflétant la

diète qui est promulgué par les holistes des systèmes d’une part, et les réductionnistes des

éléments d’autre part, il nous reste finalement les simulistes et leurs modèles.

L’identification et même la quantification du régime alimentaire par la modélisation des

relations entre les profils d’acides gras des proies et des prédateurs comme ça se fait en

QFASA est un objectif direct et ne fait pas partie des étapes ‘préliminaires’ qui sont issues

des analyses lipidiques. En disant ceci, il reste à valider en général quelques aspects du

comportement dans le processus de modélisation pour produire des résultats satisfaisants.

Plus précisement, après avoir écarté des questions d’écologie, de chimie et de statistiques, il

demeure nécessaire de compenser pour le sujet, c’est-à-dire l’espèce et la classe de

prédateur (taxonomie, maturité, sexe, ou santé), qui pourrait avoir une influence dans la

modélisation des profils en acides gras. Des exemples de catégories incluent vertébrés et

invertébrés, carnivores terrestres et marins, cétacés et pinnipèdes, alimentés en captivité ou

libres en nature.

Pour la validation des modèles incluant QFASA, on doit alors faire appel à deux

démarches, soit la comparaison avec un indicateur externe et la compensation pour mieux

refléter ce qui est connu avec cet indicateur. Comme dans le cas des isotopes stables, le

produit des modèles de profils en acides gras reste provisoire en l’absence de confirmation

des résultats en se référant à des paramètres connus comme ceux fournis par d’autres

indicateurs bien établis (e.g., traces de proies non-digérées), les observations d’alimentation

153

ou les manipulations de régime alimentaire en captivité. En ce faisant, on pourrait établir

des facteurs de calibration, afin d’ajuster les profils des prédateurs pour compenser

l’assimilation différentielle des acides gras provenant des proies consommées (Iverson et al.

2004, Thiemann et al. 2004a). La nécessité en même temps que la réussite de telles

approches de modèle ‘calibré’ est bien mise en évidence dans le cadre des études

présentement en cours chez les oiseaux et les pinnipèdes par les pionniers du laboratoire

Iverson et leur modèle QFASA. Mais, comme dans le cas de l’établissement d’une banque

de données de proies, il reste toujours à savoir à quel point la variabilité, ou plutôt la

généralisation des facteurs de calibration produits avec certains individus pourrait être

appliquée à d’autres espèces d’intérêt, notamment aux cetacés (Santos et Pierce 2003,

Krahn et al. 2004, Samuel et Worthy 2004).

Jusqu’à présent, il commence à y avoir des indications qu’il y aura des exceptions,

ou au moins des subtilités, aux généralités établies avec les phocidés juvéniles en captivité

en comparaison avec les autres scénarios (Grahl-Nielsen et al. 2003, Andersen et al. 2004,

Beck et al. 2005, Käkelä et al. 2005, Staniland et Pond 2005). Cela n’est pas tout à fait

surprenant, vu la variété souvent inattendue des stratégies physiologiques de l’exploitation

des lipides (Käkelä et Hyvärinen 1996, Pond 2000, Sprecher 2000, Hulbert et al. 2002,

Tocher 2003). La variété peut être regardée comme un obstacle initialement décourageant,

surtout pour les gestionnaires des projets de recherche, mais elle pourrait aussi présenter des

opportunités qui ne sont pas encore imaginées par les chercheurs. C’est l’exploitaton des

différences et ‘défauts’ qui propulse l’évolution, dont on est le produit, selon les idées

prémonitoires de Darwin. Néanmoins, les remarques aux détails et aux exceptions peuvent

rendre le discours sensible aux critiques au sujet du potentiel des acides gras comme

154

marqueurs trophiques en modélisation. Ce qui risque d’inutilement nuire au développement

des modèles, du côté des cyniques par leurs échecs et manque de rigueur en réplication,

autant que par les promoteurs présentant des résultats difficiles à simuler.

Un rappel de l’histoire et des débats autour de l’évolution pourrait nous servir

comme un dernier point pour conclure. Aujourd’hui, en 2005, la théorie de l’évolution est

toujours contestée dans l’éducation des élèves dans certains districts aux Etats-Unis. Ce qui

pourrait nous faire penser encore aux mots d’un des grands supporteurs de l’évolution,

Stephen Jay Gould, qui parmi plusieurs scientifiques s’intéressait aux phénomènes naturels

autant qu’à notre capacité à accepter des idées utiles à la science. En effet, leurs essais

littéraires nous aident à voir qu’il y a deux façons d’attaquer un problème quelconque :

l’approche a priori et l’approche historique. L’approche ‘en avant’ est celle qui est mieux

connue par les chercheurs, les ingénieurs et les gestionnaires des parcs naturels à la navette

spatiale parce que c’est ce qu’on fait tous les jours : identifier un problème et proposer des

causes et des réponses logiques et raisonables à tester selon l’évidence acquise. Dans cette

étude, on pouvait observer des patrons en acides gras et proposer des liens possibles. Sauf

qu’il est difficile de savoir ce qui est la solution unique, c’est-à-dire parcimonieuse. Mais,

comme se lamente le généticien Dobzhansky, « tout se comprend mieux dans le contexte de

l’évolution » (‘Nothing makes sense except in the light of evolution’). Ainsi, l’histoire, tant

dans l’évolution des espèces que dans les interactions écologiques des populations et les

activités d’un individu, pouvait nous fournir des idées auxquelles des explications a priori

sont correctes ou non, à cause du chemin pris par un organisme pour démontrer les patrons

observés. Dans cette étude, l’histoire est proposée comme ultimement responsable pour les

différences dans les patrons, mais on n’a à présent que des outils tels que QFASA, qui nous

155

offre ce qu’il voit analytiquement et propose les solutions probables. Quand la question à

répondre est bien délimitée, ce serait suffisant, mais en écologie on peut poser bien des

questions encore pour l’histoire à venir.

En août 2005, l’exploration de l’espace par les humains a fait les manchettes des

nouvelles pour ses difficultés. On a cherché des causes pour les échecs des programmes

extraterrestres et les gaffes d’aujourd’hui. Une exercise intéressant est de regarder le cas du

fameux réservoir externe de la navette spatiale. Ostensiblement, ses dimensions et

propriétés sont bien conçues pour répondre aux besoins des propulseurs de la navette. Mais

aussi il réflète l’histoire des matériaux et des transports en communs, comme le fait d’être

transportable par chemin de fer via des tunnels dont les dimensions étaient définies par les

normes européennes, avec leur origine ultime dans la taille des chemins construits par les

romains pour leurs chariots de guerre. Si vous proposez que le réservoir est conçu pour la

navette, vous avez raison autant que si vous proposez d’établir des liens entre diète et la

détection d’acides gras chez les proies et les prédateurs. Et si vous savez qu’en partie le

design réflète une contrainte historique qui suit la logique pas-à-pas, mais qui ultimement

n’a pas de sens et était peu probable à prédire, vous avez gagné un peu d’information

potentiellement utile. Je pense que Darwin et Gould auraient dû être bien curieux de savoir

quelles autres histoires on raconterait autour des lipides dans une domaine d’étude qui sort

de l’ennui d’être un simple tissu d’isolant et d’énergie, pour sortir avec les autres molécules

renommées, telles que l’ADN et les protéines, au bal des finissants dans les années à venir

(Pond 2000).

156

Bibliographie Générale Abookire, A. A., and J. F. Piatt. 2005. Oceanographic conditions structure forage fishes into

lipid-rich and lipid-poor communities in lower Cook Inlet, Alaska, USA. Marine

Ecology Progress Series 287:229-240.

Ackman, R. G. 1989. Marine biogenic lipids, fats, and oils. CRC Press, Boca Raton, FL.

Ackman, R. G. 2002. The gas chromatograph in practical analyses of common and

uncommon fatty acids for the 21st century. Analytica Chimica Acta 465:175-192.

Ackman, R. G., and C. A. Eaton. 1966. Lipids of the fin whale (Balaenoptera physalus)

from north Atlantic waters. III. Occurrence of eicosenoic and docosenoic fatty acids

in the zooplankter Meganyctiphanes norvegica (M. Sars) and their effect on whale

oil composition. Canadian Journal of Biochemistry 44:1561-1566.

Ackman, R. G., S. Epstein, and C. A. Eaton. 1971. Differences in the fatty acid

compositions of blubber fats from northwestern Atlantic finwhales (Balaenoptera

physalus) and harp seals (Pagophilus groenlandica). Comparative Biochemistry and

Physiology Part B 40:683-697.

Ackman, R. G., and S. N. Hooper. 1973. Non-methylene-interrupted fatty acids in lipids of

shallow-water marine invertebrates: a comparison of two molluscs (Littorina

littorea and Lunatia triseriata) with the sand shrimp (Crangon septemspinosa).

Comparative Biochemistry and Physiology Part B 46:153-165.

Ackman, R. G., J.-L. Sebedio, and M. I. P. Kovacs. 1980. Role of eicosenoic and

docosenoic fatty acids in freshwater and marine lipids. Marine Chemistry 9:

157-164.

Afifi, A. A., and V. Clark. 1990. Computer-aided multivariate analysis, 2nd edition.

Chapman & Hall, New York, NY.

Aitchison, J. 1986. The statistical analysis of compositional data. Chapman and Hall,

London, UK.

Andersen, S. M., C. Lydersen, O. Grahl-Nielsen, and K. M. Kovacs. 2004. Autumn diet of

harbour seals (Phoca vitulina) at Prins Karls Forland, Svalbard assessed via scat and

fatty-acid analyses. Canadian Journal of Zoology 82:1230-1245.

157

Anderson, M. J., S. D. Connell, B. M. Gillanders, C. E. Diebel, W. M. Bloms, J. E.

Saunders, and T. J. Landers. 2005. Relationship between taxonomic resolution and

spatial scales of multivariate variation. Journal of Animal Ecology 74: 636-646.

Anderson, P. K. 2001. Marine mammals in the next one hundred years: Twilight for a

Pleistocene megafauna? Journal of Mammalogy 82:623-629.

Andersson, C. A. 2000. Exploratory multivariate data analysis with applications in food

technology. PhD. Royal Veterinary and Agricultural University, Frederiksberg, DK.

52 p.

Arim, M., and D. E. Naya. 2003. Pinniped diets inferred from scats: analysis of biases in

prey occurrence. Canadian Journal of Zoology 81:67-73.

Arnould, J. P. Y., M. M. Nelson, P. D. Nichols, and W. H. Oosthuizen. 2005. Variation in

the fatty acid composition of blubber in Cape fur seals (Arctocephalus pusillus

pusillus) and the implications for dietary interpretation. Journal of Comparative

Physiology B 175:285-295.

Arts, M. T., R. G. Ackman, and B. J. Holub. 2001. "Essential fatty acids" in aquatic

ecosystems: a crucial link between diet and human health and evolution. Canadian

Journal of Fisheries and Aquatic Sciences 58:122-137.

Auel, H., M. Harjes, R. da Rocha, D. Stübing, and W. Hagen. 2002. Lipid biomarkers

indicate different ecological niches and trophic relationships of the Arctic hyperiid

amphipods Themisto abyssorum and T. libellula. Polar Biology 25: 374-383.

Baker, C. S., and P. J. Clapham. 2004. Modelling the past and future of whales and

whaling. Trends in Ecology & Evolution 19:365.

Bakshi, B. R., and U. Utojo. 1999. A common framework for the unification of neural,

chemometric and statistical modeling method. Analytica Chimica Acta 384:

227-247.

Beach, D. H., G. W. Harrington, J. L. Gellerman, H. Schlenk, and G. G. Holz Jr. 1974.

Biosynthesis of oleic acid and docosahexaenoic acid by a heterotrophic marine

dinoflagellate Crypthecodinium cohnii. Biochimica et Biophysica Acta 369:16-24.

Beaugrand, G. 2003. An overview of statistical methods applied to CPR data. Progress in

Oceanography 58:235-262.

158

Beck, C. A., W. D. Bowen, and S. J. Iverson. 2005. Blubber fatty acids reveal sex

differences in the diet of a size-dimorphic marine carnivore. Canadian Journal of

Zoology 83:377-388.

Beck, C. A., W. D. Bowen, J. McMillan, and S. J. Iverson. 2003. Sex differences in the

diving behaviour of a size-dimorphic capital breeder: the grey seal. Animal

Behaviour 66:777-789.

Beck, G. G., M. O. Hammill, and T. G. Smith. 1993. Seasonal variation in the diet of harp

seals (Phoca groenlandica) from the Gulf of St. Lawrence and western Hudson

Strait. Canadian Journal of Fisheries and Aquatic Sciences 50: 1363-1371.

Ben-David, M., and D. M. Schell. 2001. Mixing models in analyses of diet using multiple

stable isotopes: a response. Oecologia 127:180-184.

Benoît, H. P., M.-J. Abgrall, and D. P. Swain. 2003a. An assessment of the general status of

marine and diadromous fish species in the southern Gulf of St. Lawrence based on

annual bottom-trawl surveys (1971-2002). Can. Tech. Rep. Fish. Aquat. Sci. No.

2472. 183 p.

Benoît, H. P., E. Darbyson, and D. P. Swain. 2003b. An atlas of the geographic distribution

of marine fish and invertebrates in the southern Gulf of St. Lawrence based on

annual bottom-trawl surveys (1971-2002). Can. Data Rep. Fish. Aquat. Sci. No.

1112. 185 p.

Best, C. A., J. E. Cluette-Brown, M. Teruya, A. Teruya, and M. Laposata. 2003a. Red

blood cell fatty acid ethyl esters: a significant component of fatty acid ethyl esters in

the blood. Journal of Lipid Research 44:612-620.

Best, N. J., C. J. A. Bradshaw, M. A. Hindell, and P. D. Nichols. 2003b. Vertical

stratification of fatty acids in the blubber of southern elephant seals (Mirounga

leonina): implications for diet analysis. Comparative Biochemistry and Physiology

Part B: Biochemistry & Molecular Biology 134:253-263.

Birkeland, C., and P. K. Dayton. 2005. The importance in fishery management of leaving

the big ones. Trends in Ecology & Evolution 20:356-358.

Bonn, B. A. 1998. Polychlorinated dibenzo-p-dioxin and dibenzofuran concetration profiles

in sediment and fish tissue of the Willamette Basin, Oregon. Environmental Science

and Technology 32:729-735.

159

Born, E. W., H. Dahlgaard, F. F. Riget, R. Dietz, N. Øien, and T. Haug. 2002. Regional

variation of caesium-137 in minke whales Balaenoptera acutorostrata from West

Greenland, the Northeast Atlantic and the North Sea. Polar Biology 25:907-913.

Born, E. W., S. Rysgaard, G. Ehlmé, M. Sejr, M. Acquarone, and N. Levermann. 2003.

Underwater observations of foraging free-living Atlantic walruses (Odobenus

rosmarus rosmarus) and estimates of their food consumption. Polar Biology 26:

348-357.

Bourget, E. 1997. Les animaux littoraux du Saint-Laurent : guide d’identification. Les

presses de l’université Laval, Québec. 268 p.

Bowen, D., and G. D. Harrison. 1994. Offshore diet of grey seals Halichoerus grypus near

Sable Island, Canada. Marine Ecology Progress Series 158:267-274.

Bowen, W. D. 1997. Role of marine mammals in aquatic ecosystems. Marine Ecology

Progress Series 158:267-274.

Bowen, W. D. 2005. From the editor: reflections with comments on the collection and

analysis of data. Marine Mammal Science 21:178-183.

Bowen, W. D., S. L. Ellis, S. J. Iverson, and D. J. Boness. 2003. Maternal and newborn life-

history traits during periods of contrasting population trends: implications for

explaining the decline of harbour seals (Phoca vitulina), on Sable Island. Journal of

Zoology, London 261:155-163.

Bowen, W. D., and G. D. Harrison. 1996. Comparison of harbour seal diets in two inshore

habitats of Atlantic Canada. Canadian Journal of Zoology 74:125-135.

Bowen, W. D., D. Tully, D. J. Boness, B. M. Buleier, and G. J. Marshall. 2002. Prey-

dependent foraging tactics and prey profitability in a marine mammal. Marine

Ecology Progress Series 244:235-245.

Bradshaw, C. J. A., M. A. Hindell, N. J. Best, K. L. Phillips, G. Wilson, and C. M. Nichols.

2003. You are what you eat: describing the foraging ecology of southern elephant

seals (Mirounga leonina) using blubber fatty acids. Proceedings of the Royal

Society of London B 270:1283-1292.

Breiman, L., J. H. Friedman, R. A. Olshen, and C. J. Stone. 1984. Classification and

Regression Trees. Wadsworth International Group, Belmont, CA 358 p.

160

Bremer, J., and K. R. Noren. 1982. Metabolism of very long-chain monounsaturated fatty

acids (22:1) and the adaptation to their presence in the diet. Journal of Lipid

Research 23:243-256.

Brereton, R. G. 1990. Chemometrics: applications of mathematics and statistics to

laboratory systems. Ellis Horwood, Chichester, UK.

Brown, D. J., I. L. Boyd, G. C. Cripps, and P. J. Butler. 1999. Fatty acid signature analysis

from the milk of Antarctic fur seals and southern elephant seals from South

Georgia: implications for diet determination. Marine Ecology Progress Series

187:251-263.

Bruno, J. F., J. J. Stachowicz, and M. D. Bertness. 2003. Inclusion of facilitation into

ecological theory. Trends in Ecology & Evolution 18:119-125.

Bryant, R., I. L. Jones, and J. M. Hipfner. 1999. Responses to changes in prey availability

by Common Murres and Thick-billed Murres at the Gannet Islands, Labrador.

Canadian Journal of Zoology:1278-1287.

Budge, S. M., M. H. Cooper, and S. J. Iverson. 2004. Demonstration of the deposition and

modification of dietary fatty acids in pinniped blubber using radiolabelled

precursors. Physiological and Biochemical Zoology 77:682-687.

Budge, S. M., and S. J. Iverson. 2003. Quantitative analysis of fatty acid precursors in

marine samples: direct conversion of wax ester alcohols and dimethylacetals to fatty

acid methyl esters. Journal of Lipid Research 44:1802-1807.

Budge, S. M., S. J. Iverson, W. D. Bowen, and R. G. Ackman. 2002. Among- and within-

species variability in fatty acid signatures of marine fish and invertebrates on the

Scotian Shelf, Georges Bank, and southern Gulf of St. Lawrence. Canadian Journal

of Fisheries and Aquatic Sciences 59:886-898.

Budge, S. M. 1999. Fatty acid biomarkers in a cold water environment. PhD Thesis,

Memorial University, St. John’s, NF.

Bundy, A. 2001. Fishing on ecosystems: the interplay of fishing and predation in

Newfoundland-Labrador. Canadian Journal of Fisheries and Aquatic Sciences

58:1153-1167.

161

Burdge, G. C., and P. C. Calder. 2005. alpha-linolenic acid metabolism in adult humans: the

effects of gender and age on conversion to longer-chain polyunsaturated fatty acids.

European Journal of Lipid Science and Technology 107:426-439.

Burnham, K. P., and D. R. Anderson. 1999. Kullback-Leibler information as a basis for

strong inference in ecological studies. Wildlife Research 28:111-119.

Burns, J. J., S. J. Trumble, M. A. Castellini, and J. W. Testa. 1998. The diet of Weddell

seals in McMurdo Sound, Antarctica as determined from scat collections and stable

isotope analysis. Polar Biology 19:272-282.

Bérubé, S., and J.-D. Lambert. 1999. Communautés ichtyennes côtières de l'estuaire du

Saint-Laurent en 1996 et 1997: suite du suivi ichtyologique (1986-1995). Rap. tech.

can. sci. halieut. aquat. No. 2281. 62 p.

Caron, F., D. Fournier, P. Nellis, and P. Y. Collin. 2001. Biodiversité ichtyologique à la

rencontre de l'estuaire fluvial du Saint-Laurent en 2000. Societe de la faune et des

parcs du Quebec, Quebec. 61 p.

Carscadden, J. E. 2005. Did signals from seabirds indicate changes in capelin biology

during the 1990s? Comment on Davoren & Montevecchi (2003). Marine Ecology

Progress Series 285:289-297.

Castonguay, M., P. V. Hodson, C. M. Couillard, M. J. Eckersley, J.-D. Dutil, and G.

Verrault. 1994. Why is recruitment of the American eel, Anguilla rostrata, declining

in the St. Lawrence River and Gulf. Canadian Journal of Fisheries and Aquatic

Sciences 51:479-488.

Caswell, H., M. Fujiwara, and S. Brault. 1999. Declining survival probability threatens the

North Atlantic right whale. Proceedings of the National Academy of Sciences, USA

96:3308-3313.

Chamberlain, P. M., I. D. Bull, H. I. J. Black, P. Ineson, and R. P. Evershed. 2004. Lipid

content and carbon assimilation in Collembola: implications for the use of

compound-specific carbon isotope analysis in animal dietary studies. Oecologia

139:325-335.

Cherel, Y., and G. Duhamel. 2004. Antarctic jaws: cephalopod prey of sharks in Kerguelen

waters. Deep-Sea Research I 51:17-31.

162

Choi, J. S., K. T. Frank, W. C. Leggett, and K. F. Drinkwater. 2004. Transition to an

alternate state in a continental shelf ecosystem. Canadian Journal of Fisheries and

Aquatic Sciences 61:505-510.

Christensen, V., S. Guénette, J. J. Heymans, C. J. Walters, R. Watson, D. Zeller, and D.

Pauly. 2003. Hundred-year decline of North Atlantic predatory fishes. Fish and

Fisheries 4:1-24.

Christensen, V., and D. Pauly. 1992. Ecopath II--a software for balancing steady-state

ecosystem models and calculation network characteristics. Ecological Modelling

61:169-185.

Christie, W. W. 2003. Lipid Analysis: isolation, separation, identification and structural

analysis of lipids. The Oily Press, Bridgwater. 416 p.

Colbourne, E. B., E. G. Dawe, D. G. Parsons, E. F. Murphy, W. R. Bowering, E. L. Dalley,

J. T. Anderson, J. B. Dempson, D. Orr, D. E. Stansbury, and G. P. Ennis. 2002. A

preliminary review of environmental-stock relationships for some species of marine

organisms in NAFO waters of the Northwest Atlantic. in NAFO Scientific Council

Reports 02/34.

Connor, S. L., N. Zhu, G. J. Anderson, E. Jafffe, J. Carlson, and S. L. Connor. 2000. Cheek

cell phospholipids in human infants: a marker of docosahexaenoic and arachidonic

acids in the diet, plasma, and red blood cells. American Journal of Clinical Nutrition

71:21-27.

Cook, H. W. 1996. Fatty acid desaturation and chain elongation in eukaryotes. Pages 129-

152 in D. E. Vance and J. E. Vance, editors. Biochemistry of lipids, lipoproteins and

membranes. Elsevier Science B.V.

Cooper, M. H., S. J. Iverson, and H. Heras. 2005. Dynamics of blood chylomicron fatty

acids in a marine carnivore: implications for lipid metabolism and quantitative

estimation of predator diets. Journal of Comparative Physiology B 175:133-145.

COSEWIC. 2003. Annual report to the Canadian Endangered Species Conservation

Council by the Committee on the Status of Endangered Wildlife in Canada

(COSEWIC). http://www.sararegistry.gc.ca/gen_info/cosewic_annual_e.cfm

163

COSEWIC. 2004. COSEWIC assessment and update status report on the beluga whale

Delphinapterus leucas in Canada. Committee on the Status of Endangered Wildlife

in Canada, Ottawa. http://www.cosewic.gc.ca/eng/sct2/index_e.cfm

Cripps, G. C., and A. Atkinson. 2000. Fatty acid composition as an indicator of carnivory in

Antarctic krill, Euphausia superba. Canadian Journal of Fisheries and Aquatic

Sciences 57 (Suppl. 3):31-37.

Dahl, T. M., S. Falk-Petersen, G. W. Gabrielsen, J. R. Sargent, H. Hop, and R.-M. Millar.

2003. Lipids and stable isotopes in common eider, black-legged kittiwake and

northern fulmar: a trophic study from an Arctic fjord. Marine Ecology Progress

Series 256:257-269.

Dahl, T. M., C. Lydersen, K. M. Kovacs, S. Falk-Petersen, J. R. Sargent, I. Gjertz, and B.

Gulliksen. 2000. Fatty acid composition of the blubber in white whales

(Delphinapterus leucas). Polar Biology 23:401-409.

Dalsgaard, J., M. St. John, G. Kattner, D. Müller-Navarra, and W. Hagen, editors. 2003.

Fatty acid trophic markers in the pelagic marine environment. Academic Press.

Das, K., G. Lepoint, Y. Leroy, and J.-M. Bouquegneau. 2003. Marine mammals from the

southern North Sea: feeding ecology data from δ13C and δ15N measurements. Marine

Ecology Progress Series 263:287-298.

Davis, R. W., L. A. Fuiman, T. M. Williams, M. Horning, and W. Hagey. 2003.

Classification of Weddell seal dives based on 3-dimensional movements and video-

recorded observations. Marine Ecology Progress Series 264:109-122.

Davoren, G. K., and W. A. Montevecchi. 2003. Signals from seabirds indicate changing

biology of capelin stocks. Marine Ecology Progress Series 258:253-261.

Davoren, G. K., and W. A. Montevecchi. 2005. Did signals from seabirds indicate changes

in capelin biology? Reply to Carscadden (2004). Marine Ecology Progress Series

285:299-309.

Dawe, E. G., E. B. Colbourne, and K. F. Drinkwater. 2000. Environmental effects on

recruitment of short-finned squid (Illex illecebrosus). ICES Journal of Marine

Science 57:1002-1013.

164

De Guise, S., D. Martineau, P. Béland, and M. Fournier. 1995. Possible mechanisms of

action of environmental contaminants on St. Lawrence beluga whales

(Delphinapterus leucas). Environmental Health Perspectives 103 Suppl 4:73-77.

De Veaux, R. D. 1995. A guided tour of modern regression methods. Proceedings of the

Section on Physical and Engineering Sciences. (Invited talk given at the 1995 fall

FTC. Shewell Award 1995). 10 p.

De'ath, G., and K. E. Fabricius. 2000. Classification and regression trees: a powerful yet

simple technique for ecological data analysis. Ecology 81.

Deagle, B. E., D. J. Tollit, S. N. Jarman, M. Hindell, A. W. Trites, and N. J. Gales. 2005.

Molecular scatology as a tool to study diet: analysis of prey DNA in scats from

captive Steller sea lions. Molecular Ecology 14:1831-1842.

DeMaster, D. P., C. W. Fowler, S. I. Perry, and M. F. Richlen. 2001. Predation and

competition: the impact of fisheries on marine-mammal populations over the next

one hundred years. Journal of Mammalogy 82:641-651.

Dempson, J. B., M. Shears, and M. Bloom. 2001. Spatial and temporal variability in the diet

of anadromous Arctic charr, Salvelinus alpinus, in northern Labrador.

Environmental Biology of Fishes 64:49-62.

DeNiro, M. J., and S. Epstein. 1981. Influence of diet on the distribution of nitrogen

isotopes in animals. Geochimica et Cosmochimica Acta 45:341-351.

Desvilettes, C., G. Bourdier, C. Amblard, and B. Barth. 1997. Use of fatty acids for the

assessment of zooplankton grazing on bacteria, protozoans and microalgae.

Freshwater Biology 38:629-637.

deYoung, B., R. Harris, J. Alheit, G. Beaugrand, N. Mantua, and L. Shannon. 2004.

Detecting regime shifts in the ocean: data considerations. Progress in

Oceanography 60:143-164.

Dos Santos, J., I. C. Burkow, and M. Jobling. 1993. Patterns of growth and lipid deposition

in cod (Gadus morhua L.) fed natural prey and fish-based feeds. Aquaculture

110:173-189.

Drinkwater, K. F., and D. Gilbert. 2004. Hydrographic variability in the waters of the Gulf

of St. Lawrence, the Scotian Shelf and the Eastern Gulf of Maine (NAFO Subarea

4) during 1991-2000. Journal of Northwest Atlantic Fisheries Science 34:85-101.

165

Dudoit, S., J. Fridlyand, and T. P. Speed. 2000. Comparison of discrimination methods for

the classification of tumors using gene expression data. Technical Report 576,

Department of Statistics, University of California, Berkeley.

Dunham, J. S., and D. A. Duffus. 2002. Diet of gray whales (Eschrichtius robustus) in

Clayoquot Sound, British Columbia, Canada. Marine Mammal Science 18: 419-437.

Dunstan, G. A., J. K. Volkman, S. M. Barrett, J.-M. Leroi, and S. W. Jeffrey. 1993.

Essential polyunsaturated fatty acids from 14 species of diatom (Bacillariophyceae).

Phytochemistry 35:155-161.

Dunstan, G. A., J. K. Volkman, S. M. Barrett, J.-M. Leroi, and S. W. Jeffrey. 1994.

Essential polyunsaturated fatty acids from 14 species of diatom (Bacillariophyceae).

Phytochemistry 35:155-161.

Durnford, E., and F. Shahidi. 2002. Comparison of FA compositions of selected tissues of

phocid seals of Eastern Canada using one-way and multivariate techniques. JAOCS

79:1095-1102.

El-Sabh, M. I., and N. Silverberg, editors. 1990. Oceanography of a large-scale estuarine

system: the St. Lawrence. Coastal and estuarine studies, vol. 39. Springer-Verlag,

Berlin. 434 p.

Eriksson, L., E. Johansson, N. Kettaneh-Wold, and S. Wold. 2002. Multi- and Megavariate

Data Analysis: Principles and Applications. Umetrics Academy, Umea, Sweden.

527 p.

Estes, J. A., and D. O. Duggins. 1995. Sea otters and kelp forests in Alaska: generality and

variation in a community ecological paradigm. Ecological Monographs 65:75-100.

Evans, K., M. A. Hindell, and D. Thiele. 2003. Body fat and condition in sperm whales,

Physeter macrocephalus, from southern Australian waters. Comparative

Biochemistry and Physiology Part A: Molecular & Integrative Physiology 134:847-

862.

Falandysz, J., B. Wyrzykowska, L. Strandberg, T. Puzyn, B. Strandberg, and C. Rappe.

2002. Multivariate analysis of the bioaccumulation of polychlorinated biphenyls

(PCBs) in the marine pelagic food web from the southern part of the Baltic Sea,

Poland. Journal of Environmental Monitoring 4:929-941.

166

Falk-Petersen, S., W. Hagen, G. Kattner, A. Clarke, and J. R. Sargent. 2000. Lipids, trophic

relationships, and biodiversity in Arctic and Antarctic krill. Canadian Journal of

Fisheries and Aquatic Sciences 57 (Suppl. 3):178-191.

Falk-Petersen, S., T. Haug, K. T. Nilssen, A. Wold, and T. M. Dahl. 2004. Lipids and

trophic linkages in harp seal (Phoca groenlandica) from the eastern Barents Sea.

Polar Research 23:43-50.

Fanshawe, S., G. R. Van Blaricom, and A. A. Shelly. 2003. Restored top carnivores as

detriments to the performance of marine protected areas intended for fishery

sustainability: a case study red abalones and sea otters. Conservation Biology

17:273-283.

Ferdinandusse, S., S. Denis, P. A. W. Mooijer, Z. Zhang, J. K. Reddy, A. A. Spector, and

R. J. A. Wanders. 2001. Identification of the peroxisomal {beta}-oxidation enzymes

involved in the biosynthesis of docosahexaenoic acid. Journal of Lipid Research

42:1987-1995.

Field, A. 2000. Discovering Statistics using SPSS for Windows. SAGE Publications Ltd,

London, UK. 496 p.

Flaaten, O. 1998. On the bioeconomics of predator and prey fishing. Fisheries Research

37:171-191.

Folch, J. M., M. Lees, and G. H. Sloane. 1957. A simple method for the isolation and

purification of total lipids from animal tissues. Journal of Biological Chemistry

226:497-509.

Folkow, L. P., and A. S. Blix. 1999. Diving behaviour of hooded seals (Cystophora

cristata) in the Greenland and Norwegian Seas. Polar Biology 22:61-74.

Fowler-Walker, M. J., and S. D. Connell. 2002. Opposing states of subtidal habitat across

temperate Australia: consistency and predictability in kelp canopy-benthic

associations. Marine Ecology Progress Series 240.

Frank, I. E., and S. Lanteri. 1989. Classification models: Discriminant analysis, SIMCA,

CART. Chemometrics and Intelligent Laboratory Systems 5:247-256.

Frank, K. T., B. Petrie, J. S. Choi, and W. C. Leggett. 2005. Trophic cascades in a formerly

cod-dominated ecosystem. Science 308:1621-1623.

167

Fraser, A. J., J. R. Sargent, J. C. Gamble, and D. D. Seaton. 1989. Formation and transfer of

fatty acids in an enclosed marine food chain comprising phytoplankton, zooplankton

and herring (Clupea harengus L.) larvae. Marine Chemistry 27:1-18.

Freites, L., M. J. Fernandez-Reiriz, and U. Labarta. 2002. Fatty acid profiles of Mytilus

galloprovincialis (Lmk) mussel of subtidal and rocky shore origin. Comparative

Biochemistry and Physiology B-Biochemistry & Molecular Biology 132:453-461.

Fritz, L. W., and S. Hinckley. 2005. A critical review of the regime shift-"junk food"-

nutritional stress hypothesis for the decline of the western stock of Steller sea lion.

Marine Mammal Science 21:476-518.

Gallagher, R., and T. Appenzeller. 1999. Beyond reductionism. Science 284:79.

Gannes, L. Z., C. Martinez del Rio, and P. Koch. 1998. Natural abundance variations in

stable isotopes and their potential uses in animal physiological ecology.

Comparative Biochemistry and Physiology Part A: Molecular & Integrative

Physiology 119:725-737.

Gaskin, D. E. 1982. Diet and feeding behaviour of Cetacea. Pp: 31-37 in The ecology of

whales and dolphins. Heinemann, London.

Gilbert, D., P. S. Galbraith, C. Lafleur, and B. Pettigrew. 2004. Physical oceanographic

conditions in the Gulf of St. Lawrence in 2003. Can. Sci. Adv. Sec. Res. Doc.

2004/061. 63 p.

Gosselin, J. F., V. Lesage, and A. Robillard. 2001. Population index estimate for the beluga

of the St. Lawrence River. Can. Sci. Adv. Sec. Res. Doc. 2001/049. 21 p.

Graeve, M., G. Kattner, and D. Peipenburg. 1997. Lipids in Arctic benthos: does the fatty

acid and alcohol composition reflect feeding and trophic interactions? Polar Biology

18:53-61.

Grahl-Nielsen, O., M. Andersen, A. E. Derocher, C. Lydersen, O. Wiig, and K. M. Kovacs.

2003. Fatty acid composition of the adipose tissue of polar bears and of their prey:

ringed seals, bearded seals and harp seals. Marine Ecology Progress Series

265:275-282.

Grahl-Nielsen, O., M. Andersen, A. E. Derocher, C. Lydersen, O. Wiig, and K. M. Kovacs.

2004. Reply to Comment on Grahl-Nielsen et al. (2003): sampling, data treatment

168

and predictions in investigations on fatty acids in marine mammals. Marine

Ecology Progress Series 281:303-306.

Greene, C. H., and A. J. Pershing. 2004. Climate and conservation biology of North

Atlantic Right whales: being a right whale at the wrong time? Frontiers in Ecology

and the Environment 2:29-34.

Greene, D. H. S., and D. P. Selivonchick. 1987. Lipid metabolism in fish. Progress in Lipid

Research 26:53-85.

Hair, J. F., R. E. Anderson, R. L. Tatham, and W. C. Black. 1998. Multivariate Data

Analyis, 5th edition. Prentice-Hall. 730 p.

Hammill, M. O., J. F. Gosselin, F. Proust, and D. Chabot. 1999. A review of information

relevant to the issue of consumption of Atlantic cod (Gadus morhua) by seal species

in the southern Gulf of S. Lawrence. Can. Stock Ass. Secr. Res. Doc. 99/70. 26 p.

Hammill, M. O., C. Lydersen, K. M. Kovacs, and B. Sjare. 1995. Estimated fish

consumption by hooded seals (Cystophora cristata), in the Gulf of St. Lawrence.

Journal of Northwest Atlantic Fishery Science 22:249-257.

Hammill, M. O., and G. B. Stenson. 2000. Estimated prey consumption by harp seals

(Phoca groenlandica), hooded seals (Cystophora cristata), grey seals (Halichoerus

grypus), and Harbour seals (Phoca vitulina), in Atlantic Canada. Journal of

Northwest Atlantic Fishery Science 26:1-23.

Hanson, J. M., and G. A. Chouinard. 2002. Diet of Atlantic cod in the southern Gulf of St.

Lawrence as an index of ecosystem change, 1959-2000. Journal of Fish Biology

60:902-922.

Harrison, G., E. Colbourne, D. Gilbert, and B. Petrie. 2004. Oceanographic observations

and data products derived from large-scale fisheries resource assessment and

environmental surveys in the Atlantic zone. AZMP Bulletin PMZA 4:17-23.

Harvey, J. T., and G. A. Antonelis. 1994. Biases associated with non-lethal methods of

determining the diet of northern elephant seals. Marine Mammal Science 10:

178-187.

Harvey, M., J.-F. St.-Pierre, L. Devine, A. Gagné, Y. Gagnon, and M. F. Beaulieu. 2004.

Oceanographic conditions in the Estuary and the Gulf of St. Lawrence during 2003:

zooplankton. Can. Sci. Adv. Sec. Res. Doc. 2004/080. 31 p.

169

Hastie, T., R. Tibshirani, and J. H. Friedman. 2001. The elements of statistical learning:

data mining, inference, and prediction. Springer, New York. 533 p.

Hauksson, E., and V. Bogason. 1997. Comparative feeding of grey (Halichoerus grypus)

and common seals (Phoca vitulina) in coastal waters of iceland, with a note on the

diet of hooded (Cystophora cristata) and harp seals (Phoca groenlandica). Journal

of Northwest Atlantic Fisheries Science 22:125-135.

Hedd, A., R. Gales, and D. Renouf. 1996. Can stomach temperature telemetry be used to

quantify prey consumption by seals? Polar Biology 16:261-270.

Henderson, R. J., J. R. Sargent, and C. C. E. Hopkins. 1984. Changes in the content and

fatty acid composition of lipid in an isolated population of the capelin Mallotus

villosus during sexual maturation and spawning. Marine Biology 78:255-263.

Hickie, B. E., M. C. S. Kingsley, P. V. Hodson, D. C. G. Muir, P. Béland, and D. Mackay.

2000. A modelling-based perspective on the past, present, and future

polychlorinated biphenyl contamination of the St. Lawrence beluga whale

(Delphinapterus leucas) population. Canadian Journal of Fisheries and Aquatic

Sciences 57(S1):101-112.

Hill, C., M. A. Quigley, J. F. Cavaletto, and W. Gordon. 1992. Seasonal changes in lipid

content and composition in the benthic amphipods Monoporeia affinis and

Pontoporeia femorata. Limnology and Oceanography 37:1280-1289.

Hjelset, A. M., M. Andersen, I. Gjertz, C. Lydersen, and B. Gulliksen. 1999. Feeding habits

of bearded seals (Erignathus barbatus) from the Svalbard area, Norway. Polar

Biology 21:186-193.

Hjermann, D. Ø., N. C. Stenseth, and G. Ottersen. 2004. Indirect climate forcing of the

Barents Sea capelin: a cohort effect. Marine Ecology Progress Series 273: 229-238.

Hobbs, K. E., D. C. G. Muir, E. W. Born, R. Dietz, T. Haug, T. Metcalfe, C. Metcalfe, and

N. Øien. 2003. Levels and patterns of persistent organochlorines in minke whale

(Balaenoptera acutorostrata) stocks from the North Atlantic and European Arctic.

Environmental Pollution 121:239-252.

Hobbs, N. T., and R. Hilborn. 2006. Alternatives to statistical hypothesis testing in ecology:

a guide to self teaching. Ecological Applications 16:5-19.

170

Hobson, K. A. 1999. Tracing origins and migration of wildlife using stable isotopes: a

review. Oecologia 120:314-326.

Holland, D. L., J. Davenport, and J. East. 1990. The fatty acid composition of the

leatherback turtle Dermochelys coriacea and its jellyfish prey. Journal of the

Marine Biological Association of the United Kingdom 70:761-770.

Hooker, S. K., S. J. Iverson, P. Orstrom, and S. C. Smith. 2001. Diet of northern

bottlenoose whales inferred from fatty-acid and stable isotope analyses of biopsy

samples. Canadian Journal of Zoology 79:1442-1454.

Howell, K. L., D. S. M. Billett, P. A. Tyler, and R. Davidson. 2004. Feeding ecology of

deep-sea seastars (Echinodermata: Asteroidea): a pigment biomarker approach.

Marine Ecology Progress Series 266:103-110.

Howell, K. L., D. W. Pond, D. S. M. Billett, and P. A. Tyler. 2003. Feeding ecology of

deep-sea seastars (Echinodermata: Asteroidea): a fatty-acid biomarker approach.

Marine Ecology Progress Series 255:193-206.

Hulbert, A. J., T. Rana, and P. Couture. 2002. The acyl composition of mammalian

phospholipids: an allometric analysis. Comparative Biochemistry and Physiology

Part B: Biochemistry and Molecular Biology 132:515-527.

Huntington, H. P. 2000. Traditional knowledge of the ecology of beluga Delphinapterus

leucas, Cook Inlet, Alaska. Marine Fisheries Review 62:134-140.

Imfeld, S. 2000. Time, points and space: towards a better analysis of wildlife data in GIS.

PhD. University of Zurich. 145 p.

Iverson, S. J. 1993. Milk secretion in marine mammals in relation to foraging: can milk

fatty acids predict diet? Symposium of the Zoological Society of London 66:

263-291.

Iverson, S. J., J. P. Y. Arnould, and I. L. Boyd. 1997a. Milk fatty acid signatures indicate

both major and minor shifts in the diet of lactating Antarctic fur seals. Canadian

Journal of Zoology 75:188-197.

Iverson, S. J., C. Field, W. D. Bowen, and W. Blanchard. 2004. Quantitative fatty acid

signature analysis: a new method of estimating predator diets. Ecological

Monographs 74:211-235.

171

Iverson, S. J., K. J. Frost, and S. L. C. Lang. 2002. Fat content and fatty acid composition of

forage fish and invertebrates in Prince William Sound, Alaska: factors contributing

to among and within species variability. Marine Ecology Progress Series 241:161-

181.

Iverson, S. J., K. J. Frost, and L. F. Lowry. 1997b. Fatty acid signatures reveal fine scale

structure of foraging distribution of harbour seals and prey in Prince William Sound,

Alaska. Marine Ecology Progress Series 151:255-271.

James, F. C., and C. E. McCulloch. 1990. Multivariate analysis in ecology and systematics:

panacea or Pandora's box? Annual Review of Ecology and Systematics 21:129-166.

Jamieson, G. R., and E. H. Reid. 1972. The component fatty acids of some marine algal

lipids. Phytochemistry 11:1423-1432.

Jeffs, A. G., P. D. Nichols, B. D. Mooney, K. L. Phillips, and C. F. Phleger. 2004.

Identifying potential prey of the pelagic larvae of the spiny lobster Jasus edwardsii

using signature lipids. Comparative Biochemistry and Physiology B: Biochemistry

& Molecular Biology 137:487-507.

Jennings, S., K. J. Warr, and S. Mackinson. 2002. Use of size-based production and stable

isotope analyses to predict trophic transfer efficiencies and predator-prey body mass

ratios in food webs. Marine Ecology Progress Series 240:11-20.

Joensen, H., P. Steingrund, I. Fjallstein, and O. Grahl-Nielsen. 2000. Discrimination

between two reared stocks of cod (Gadus morhua) from the Faroe Islands by

chemometry of the fatty acid composition in the heart tissue. Marine Biology

136:573-580.

Kaiser, H. F. 1958. The varimax criterion for analytic rotation in factor analysis.

Psychometrika 23:187-200.

Kanazawa, A., S. Teshima, and K. Ono. 1979. Relationship between essential fatty acid

requirements of aquatic animals and the capacity of bioconversion of linolenic acid

to highly unsaturated fatty acids. Comparative Biochemistry and Physiology

63B:295-298.

Kapel, F. O. 2000. Feeding habits of harp and hooded seals in Greenland waters. Pp. 50-64

in G. A. Vikingsson and F. O. Kapel, editors. Minke Whales, Harp and Hooded

172

Seals: Major Predators in the North Atlantic Ecosystem. NAMMCO Scientific

Publications, Tromsø.

Kattner, G., and M. Krause. 1987. Changes in lipids during the development of Calanus

finmarchicus s.l. from Copepodid I to adult. Marine Biology 96:511-518.

Kenney, R. D., G. P. Scott, T. J. Thompson, and H. E. Winn. 1997. Estimates of Prey

Consumption and Trophic Impacts of Cetaceans in the USA Northeast Continental

Shelf Ecosystem. Journal of Northwest Atlantic Fisheries Science 22:155-171.

Kingsley, M. C. S. 2002. Status of the belugas of the St. Lawrence estuary, Canada. Pp.

239-257 in M. P. Heide-Jørgensen and Ø. Wiig, editors. Belugas in the North

Atlantic and the Russian Arctic. NAMMCO Scientific Publications, Tromsø.

Kingsley, M. C. S., and R. R. Reeves. 1998. Aerial surveys of cetaceans in the Gulf of St.

Lawrence in 1995 and 1996. Canadian Journal of Zoology 76:1529-1550.

Kinze, C. C. 2003. Marine Mammals of the North Atlantic. Princeton University Press.

Kirsch, P. E., S. J. Iverson, and W. D. Bowen. 2000. Effect of a low-fat diet on body

composition and blubber fatty acids of captive juvenile harp seals (Phoca

groenlandica). Physiological and Biochemical Zoology 73:45-59.

Kirsch, P. E., S. J. Iverson, W. D. Bowen, S. R. Kerr, and R. G. Ackman. 1998. Dietary

effects on the fatty acid signature of whole Atlantic cod (Gadus morhua). Canadian

Journal of Fisheries and Aquatic Sciences 55:1378-1386.

Kitts, D. D., A. W. Trites, M. D. Hunyh, C. Hu, and A. W. Trites. 2004. Season variation in

nutrient composition of Alaskan walleye pollock. Canadian Journal of Zoology

82:1408-1415.

Koopman, H. N. 2001. The structure and function of the blubber of odontocetes. PhD. Duke

University, Durham, NC. 425 p.

Koopman, H. N., D. A. Pabst, W. A. McLellan, R. M. Dillaman, and A. J. Read. 2002.

Changes in blubber distribution and morphology associated with starvation in the

harbor porpoise (Phocoena phocoena): evidence for regional differences in blubber

structure and function. Physiological and Biochemical Zoology 75: 498-512.

Krahn, M. M., D. P. Herman, G. M. Ylitalo, C. A. Sloan, D. G. Burrows, R. C. Hobbs, B.

A. Mahoney, G. K. Yanagida, J. Calambokidis, and S. E. Moore. 2004.

Stratification of lipids, fatty acids and organochlorine contaminants in blubber of

173

white whales and killer whales. Journal of Cetacean Research and Management

62:175-189.

Krahn, M. M., G. M. Ylitalo, J. E. Stein, A. Aguilar, and A. Borrell. 2003. Organochlorine

contaminants in cetaceans: how to facilitate interpretation and avoid errors when

comparing datasets. Journal of Cetacean Research and Management 5:103-113.

Käkelä, R., and H. Hyvärinen. 1996. Site-specific fatty acid composition in adipose tissues

of several northern aquatic and terrestrial mammals. Comparative Biochemistry and

Physiology Part B 115:501-514.

Käkelä, R., and H. Hyvärinen. 1998. Composition of polyunsaturated fatty acids in the liver

of freshwater and marine ringed seals (Phoca hispida ssp.) differs largely due to the

diet of the seals. Comparative Biochemistry and Physiology Part B 120:231-237.

Käkelä, R., H. Hyvärinen, and P. Vainiotalo. 1993. Fatty acid composition in liver and

blubber of the Saimaa ringed seal (Phoca hispida saimensis) compared with that of

the ringed seal (Phoca hispida botnica) and grey seal (Halichoerus grypus) from the

Baltic. Comparative Biochemistry and Physiology Part B 105B:553-565.

Käkelä, R., A. Käkelä, S. Kahle, P. H. Becker, A. Kelly, and R. W. Furness. 2005. Fatty

acid signatures in plasma of captive herring gulls as indicative of dermersal or

pelagic fish diet. Marine Ecology Progress Series 293:191-200.

Laidre, K. L., and M. P. Heide-Jørgensen. 2005. Winter feeding intensity of narwhals

(Monodon monceros). Marine Mammal Science 21:45-57.

Lapillonne, A., J. C. DeMar, V. Nannegari, and W. C. Heird. 2002. The fatty acid profile of

buccal cheek cell phospholipids is a noninvasive marker of long-chain

polyunsaturated fatty acids status in piglets. Journal of Nutrition 132:2319-2323.

Laurin, J. 1982. Étude écologique et éthologique de la population de bélugas

(Delphinapterus leucas) du fjord du Saguenay, Québec. M. A. Université de

Montréal, Montreal, QC. 145 p.

Lavigueur, L., M. O. Hammill, and S. Asselin. 1993. Distribution et biologie des phoques et

autres mammifères marins dans la région du parc marin du Saguenay, Rapp.

Manuscrit Can. Sci. Halieut. Aquat. No. 2220. 40 p.

174

Law, R. J., M. Alaee, C. R. Allchin, J. P. Boon, M. Lebeuf, P. Lepom, and G. A. Stern.

2003. Levels and trends of polybrominated diphenylethers and other brominated

flame retardants in wildlife. Environment International 29:757-770.

Lawson, J. W., J. T. Anderson, E. L. Dalley, and G. B. Stenson. 1998a. Selective foraging

by harp seals Phoca groenlandica in nearshore and offshore waters of

Newfoundland, 1993 and 1994. Marine Ecology Progress Series 163:1-10.

Lawson, J. W., A. M. Magalhaes, and E. H. Miller. 1998b. Important prey species of

marine vertebrate predators in the Northwest Atlantic: proximate composition and

energy density. Marine Ecology Progress Series 164:13-20.

Lea, M.-A., Y. Cherel, C. Guinet, and C. M. Nichols. 2002a. Antarctic fur seals foraging in

the Polar Frontal Zone: inter-annual shifts in diet as shown from fecal and fatty acid

analyses. Marine Ecology Progress Series 245:281-297.

Lea, M.-A., P. D. Nichols, and G. Wilson. 2002b. Fatty acid composition of lipid-rich

myctophids and mackeral icefish (Champsocephalus gunnari) - Southern Ocean

food-web implications. Polar Biology 25:843-854.

Legendre, L., and P. Legendre. 1998. Numerical Ecology, Second English edition. Elsevier

Science B. V., Amsterdam. 853 p.

Legendre, P., K. E. Ellingsen, E. Bjornbom, and P. Casgrain. 2002. Acoustic seabed

classification: improved statistical method. Canadian Journal of Fisheries and

Aquatic Sciences 59:1-5.

Leonard, A. E., S. L. Pereira, H. Sprecher, and Y.-S. Huang. 2004. Elongation of long-

chain fatty acids. Progress in Lipid Research 43:36-54.

Lesage, V. 2004. Long-distance movements of harbour seals (Phoca vitulina) from a

seasonally ice-covered area, the St. Lawrence River estuary, Canada. Canadian

Journal of Zoology 82:1070-1081.

Lesage, V., M. O. Hammill, and K. M. Kovacs. 2001. Marine mammals and the community

structure of the Estuary and Gulf of St. Lawrence, Canada: evidence from stable

isotope analysis. Marine Ecology Progress Series 210:203-221.

Lesage, V., and M. C. S. Kingsley. 1998. Updated status of the St. Lawrence River

population of the beluga, Delphinapterus leucas. Canadian Field-Naturalist 112:98-

114.

175

Levermann, N., A. Galatius, G. Ehime, S. Rysgaard, and E. W. Born. 2003. Feeding

behaviour of free-ranging walruses with notes on apparent dextrality of flipper use.

BMC Ecology 3 (9).

Lin, D. S., and W. E. Conner. 1990. Are the n-3 fatty acids from dietary fish oil deposited

in the triglyceride stores of adipose tissue? American Journal of Clinical Nutrition

51:535-539.

Litzow, M. A. 2004. Spatiotemporal predictability of schooling and nonschooling prey of

pigeon guillemots. The Condor 106:410-415.

Lydersen, C., I. Gjertz, and J. M. Weslawski. 1989. Stomach contents of autumn-feeding

marine vertebrates from Hornsund, Svalbard. Polar Record 25:107-114.

MacKenzie, B. R., J. Alheit, D. J. Conley, P. Holm, and C. C. Kinze. 2002. Ecological

hypotheses for a historical reconstruction of upper trophic level biomass in the

Baltic Sea and Skagerrak. Canadian Journal of Fisheries and Aquatic Sciences

59:173-190.

Mansour, M. P., D. G. Holdsworth, S. E. Forbes, C. K. Macleod, and J. K. Volkman. 2005.

High contents of 24:6(n-3) and 20:1(n-13) fatt acids in the brittle star Amphiura

elandiformis from Tasmanian coastal sediments. Biochemical Systematics and

Ecology 33:659-674.

Marangoni, F., C. Colombo, and C. Galli. 2004. A method for the direct evaluation of the

fatty acid status in a drop of blood from a fingertip in humans: applicability to

nutritional and epidemiological studies. Analytical Biochemistry 326:267-272.

Marengo, E., M. Aceto, and V. Maurino. 2001. Classification of Nebbiolo-based wines

from Piedmont (Italy) by means of solid-phase microextraction-gas

chromatography-mass spectrometry of volatile compounds. Journal of

Chromatography A 943:123-137.

Martin, A. R., and T. G. Smith. 1999. Strategy and capability of wild belugas,

Delphinapterus leucas, during deep, benthic diving. Canadian Journal of Zoology

77:1783-1793.

Martin-Fernandez, J. A., and C. Barcelo-Vidal. 1998. A Critical Approach to Non-

Parametric Classification of Compositional Data. Pp. 49-56 in A. Rizzi, editor.

Advances In Data Science And Classification: Proceedings Of The 6th Conference

176

Of The International Federation Of Classification Societies (IFCS-98). Springer-

Verlag Berlin and Heidelberg GmbH & Co. KG.

Martineau, D., S. De Guise, M. Fournier, L. Shugart, C. Girard, A. Lagacé, and P. Béland.

1994. Pathology and toxicology of beluga whales from the St. Lawrence Estuary,

Qubec, Canada. Past, present, and future. The Science of the Total Environment

154:210-215.

Mau, T. L. 2004. Investigations of the role of lipids in marine mammal diets, health and

ecology. PhD. University of Alaska Fairbanks. 146 p.

May, R. 1999. Unanswered questions in ecology. Philosophical Transactions of the Royal

Society of London: B 354:1951-1959.

McClean, S. A., M. A. Rumble, R. M. King, and W. L. Baker. 1998. Evaluation of resource

selection methods with different definitions of availability. Journal of Wildlife

Management 62:793-801.

McIntyre, S., and R. Mckitrick. 2005. Hockey sticks, principal components, and spurious

signficance. Geophysical Research Letters 32:L03710.

McKenzie, D. J. 2000. Effects of dietary fatty acid composition on metabolic rate and

responses to hypoxia in the European eel (Anguilla anguilla). Fish Physiology and

Biochemistry 22:281-296.

McLaren, I. A., S. Brault, J. Harwood, and D. Vardy. 2001. Report of the Eminent Panel of

Seal Management. Communications Branch, Fisheries and Oceans Canada, Ottawa.

DFO/6201. 143 p.

Mercer, J. A. 1970. Sur la limite septrionale du calmar Loligo pealei (LeSueur). Le

Naturaliste Canadien 97:823-824.

Merrick, R. L., M. K. Chumbley, and G. V. Byrd. 1997. Diet diversity of Steller sea lions

(Eumetopias jubatus) and their population decline in Alaska: a potential

relationship. Canadian Journal of Fisheries and Aquatic Sciences 54:1342-1348.

Michener, R. H., and D. M. Schell. 1994. Stable isotope ratios as tracers in marine aquatic

food webs. Pp. 138-157 in K. Lajtha and R. H. Michener, editors. Stable Isotopes in

Ecology and Environmental Science. Methods in Ecology. Blackwell, Boston.

177

Minigawa, M., and E. Wada. 1984. Stepwise enrichment of 15N along food chains: Further

evidence and the relation between δ15N and animal age. Geochimica et

Cosmochimica Acta 48:1135-1140.

Mitchell, E. D. 1974. Present status of northwest Atlantic fin and other whale stocks. Pp.

108-169 in W. E. Schevill, editor. The Whale Problem: a status report. Harvard

University Press, Cambridge, Mass.

Morais, S., D. Boaventura, L. Narciso, P. Ré, and S. J. Hawkins. 2003. Gonad development

and fatty acid composition of Patella depressa Pennant (Gastropoda:

Prosobranchia) populations with different patterns of spatial distribution, in

exposed and sheltered sites. Journal of Experimental Marine Biology and Ecology

294:61-80.

Mottram, H. R., S. E. Woodbury, J. B. Rossell, and R. P. Evershed. 2003. High-resolution

detection of adulteration of maize oil using multi-component compound-specific

δ13C values of major and minor components and discriminant analysis. Rapid

Communications in Mass Spectrometry 17:706-712.

Muir, D. C. G., C. A. Ford, and R. E. A. Stewart. 1990. Organochlorine contaminants in

belugas, Delphinapterus leucas, from Canadian waters. Pp. 165-190 in Advances in

research on the beluga whale, Delphinapterus leucas, Ottawa.

Munro, D., D. W. Thomas, and M. M. Humphries. 2005. Torpor patterns of hibernating

eastern chipmunks. Journal of Animal Ecology 74:692-700.

Myers, R. A., J. A. Hutchings, and N. J. Barrowman. 1997. Why do fish stocks collapse?

The example of cod in Atlantic Canada. Ecological Applications 7:91-106.

Myers, R. A., and B. Worm. 2003. Rapid worldwide depletion of predatory fish

communities. Nature 423:280-283.

Møller, P., E. W. Born, R. Dietz, T. Haug, D. E. Ruzzante, and N. Øien. 2003. Regional

differences in fatty acid composition in common minke whales (Balaenoptera

acutorostrata) from the North Atlantic. Journal of Cetacean Research and

Management 5:115-124.

Nakamura, M. T., and T. Y. Nara. 2003. Essential fatty acid synthesis and its regulation in

mammals. Prostaglandins, Leukotrienes and Essential Fatty Acids 68:145-150.

178

Nichol, L. M., E. J. Gregr, R. Flinn, J. K. B. Ford, R. Gurney, L. Michaluk, and A. Peacock.

2002. British Columbia commercial whaling catch data 1908 to 1967: a detailed

description of the B.C. historical whaling database. Can. Tech. Rep. Fish. Aquat.

Sci. No. 2396. 77 p.

Nilssen, K. T., T. Haug, V. Potelov, and Y. K. Timoshenko. 1995. Feeding habits of harp

seals (Phoca groenlandica) during early summer and autumn in the northern

Barents Sea. Polar Biology 15:485-493.

Nozères, C., M. Bérubé, V. Haeberlé, R. Miller, and F. Proust. 2003. Marine Species

Identification Guide for the St. Lawrence. Maurice Lamontagne Institute, Fisheries

and Oceans Canada. Fs23-423/2003-MRC.

Ohman, M. D. 1996. Freezing and storage of copepod samples for the analysis of lipids.

Marine Ecology Progress Series 130:295-298.

Olsen, E., and O. Grahl-Nielsen. 2003. Blubber fatty acids of minke whales: stratification,

population identification and relation to diet. Marine Biology 142:13-24.

Orr, A. J., A. S. Banks, S. Mellman, H. R. Huber, R. L. DeLong, and R. F. Brown. 2004.

Examination of the foraging habits of Pacific harbor seal (Phoca vitulina richardsi)

to describe their use of the Umpqua River, Oregon and their predation on salmonids.

Fishery Bulletin 102:108-117.

Osborne, J. W., and A. B. Costello. 2004. Sample size and subject to item ratio in principal

components analysis. Practical Assessment, Research, and Evaluation 9.

Pauly, D., V. V. Christensen, J. Dalsgaard, R. Froese, and F. Torres, Jr. 1998a. Fishing

down marine food webs. Science 279:860-863.

Pauly, D., and J. Maclean. 2003. In a perfect ocean: the state of fisheries and ecosystems in

the North Atlantic Ocean. Island Press, Washington. 175 p.

Pauly, D., A. W. Trites, E. Capuli, and V. Christensen. 1998b. Diet composition and trophic

levels of marine mammals. ICES Journal of Marine Science 55:467-481.

Pawlosky, R. J., J. R. Hibbeln, Y.-H. Lin, S. Goodson, P. Riggs, N. Sebring, G. L. Brown,

and N. Salem. 2003. Effect of beef- and fish-based diets on the kinetics of n-3 fatty

acid metabolism in human subjects. American Journal of Clinical Nutrition 77:565-

572.

179

Payne, S. A., B. A. Johnston, and R. S. Otto. 1999. Proximate composition of some north-

eastern Pacific forage fish species. Fisheries Oceanography 8:159-177.

Perryman, W. L., M. A. Donahue, P. C. Perkins, and S. B. Reilly. 2002. Gray whale calf

production 1994-2000: are observed fluctuations related to changes in seasonal ice

cover. Marine Mammal Science 18:121-144.

Phillips, D. L., and J. W. Gregg. 2001. Uncertainty in source partitioning using stable

isotopes. Oecologia 127:171-179.

Phillips, K. L., P. D. Nichols, and G. D. Jackson. 2002. Lipid and fatty acid composition of

the mantle and digestive gland of four Southern Ocean squid species: implications

for food-web studies. Antarctic Science 14:212-220.

Phillips, K. L., P. D. Nichols, and G. D. Jackson. 2003. Size-related dietary changes in the

squid Moroteuthis ingens at the Falkland Islands: stomach contents and fatty-acid

analyses. Polar Biology 26:474-485.

Pierce, B. J., S. R. McWilliams, A. R. Place, and H. A. Huguenin. 2004. Diet preferences

for specific fatty acids and their effect on composition of fat reserves in migratory

Red-eyed Vireos (Vireo olivaceous). Comparative Biochemistry and Physiology

Part A 138:405-542.

Pierce, G. J., and P. R. Boyle. 1991. A review of methods for diet analysis in piscivorous

marine mammals. Oceanography Marine Biology Annual Review 29:409-486.

Pierce, R. C., D. M. Whittle, and J. B. Bramwell. 1999. Chemical Contaminants in

Canadian Aquatic Ecosystems. Department of Fisheries and Oceans, Ottawa.

Pippard, L., and H. Malcolm. 1978. White whales (Delphinapterus leucas): Observations

on their distribution and critical habitats in the St. Lawrence and Saguenay rivers.

Manuscript. Department of Indian and Northern Affairs, Parks Canada, Ottawa. 159

p.

Pitcher, K. W. 1980. Stomach contents and feces as indicators of harbour seal, Phoca

vitulina, foods in the Gulf of Alaska. Fishery Bulletin 78:797-798.

Pitcher, T. J. 2001. Fisheries managed to rebuild ecosystems? Reconstructing the past to

salvage the future. Ecological Application 11:601-617.

180

Pitcher, T. J., editor. 2004. Back to the Future: advances in methodology for modelling and

evaluating past ecosystems as future policy goals. Fisheries Centre Research

Reports No. 12. UBC, Vancouver. 162 p.

Pond, C. M. 2000. Adipose tissue, the anatomists’ Cinderella, goes to the ball at last, and

meets some influential partners. Postgraduate Medical Journal 76:671-673.

Potelov, V., K. T. Nilssen, V. Svetochev, and T. Haug. 2000. Feeding habits of harp (Phoca

groenlandica) and hooded seals (Cystophora cristata) during late winter, spring

and early summer in the Greenland Sea. in G. A. Vikingsson and F. O. Kapel,

editors. Minke Whales, Harp and Hooded Seals: Major Predators in the North

Atlantic Ecosystem. NAMMCO Scientific Publications 2:40-49.

Power, M. E. 2001. Field biology, food web models, and management: challenges of

context and scale. Oikos 94:118-129.

Préfontaine, G. 1930. Notes sur Globicephala melaena (Traill). Pp 49-66 in Rapport

Annuel de la société Provancher d’histoire naturelle canadienne.

Préfontaine, G. 1932. Notes préliminaires sur la faune de l'estuaire du Saint-Laurent dans

la région de Trois-Pistoles. Le Naturaliste Canadien 59:213-219.

Rapoport, S. I. 2003. In vivo approaches to quantifying and imaging brain arachidonic and

docosahexaenoic acid metabolism. Journal of Pediatrics 143(4 Suppl): S26-34.

Read, A. J., and C. R. Brownstein. 2003. Considering other consumers: fisheries, predators,

and Atlantic herring in the Gulf of Maine. Conservation Ecology 7:2.

Reed, J. Z., D. J. Tollit, P. M. Thompson, and W. Amos. 1997. Molecular scatology: the

use of molecular genetic analysis to assign species, sex and individual identity to

seal faece. Molecular Ecology 6:225-234.

Reeves, R. R., J. M. Breiwick, and E. D. Mitchell. 1999. History of whaling and estimated

kill of right whales, Balaena glacialis, in the northeastern United States, 1620-1924.

Marine Fisheries Review 61:1-36.

Reeves, R. R., and E. Mitchell. 1987. Cetaceans of Canada. Underwater world No. 59,

Dept. of Fisheries and Oceans. 28 p.

Reeves, R. R., and E. D. Mitchell. 1984. Catch history and initial population of white

whales (Delphinapterus leucas) in the river and Gulf of St. Lawrence, eastern

Canada. Naturaliste canadien 111:63-121.

181

Roberge, J.-M., and P. Angelstam. 2004. Usefulness of the umbrella species concept as a

conservation tool. Conservation Biology 18:76-85.

Roman, J., and S. R. Palumbi. 2003. Whales before whaling in the North Atlantic. Science

301:508-510.

Rosen, D. A. S., and A. W. Trites. 2005. Examining the potential for nutritional stress in

young Steller's sea lions: physiological effects of prey composition. Journal of

Comparative Physiology B 175.

Samuel, A. M., and G. A. J. Worthy. 2004. Variability in fatty acid composition of

bottlenose dolphin (Tursiops truncatus) blubber as a function of body site, season,

and reproductive state. Canadian Journal of Zoology 82:1933-1942.

Santos, M. B., and G. J. Pierce. 2003. The diet of harbour porpoise (Phocoena phocoena) in

the Northeast Atlantic. Oceanography Marine Biology Annual Review 41: 355-390.

Santos, R. A., and M. Haimovici. 2001. Cephalopods in the diet of marine mammals

stranded or incidentally caught along southeastern and southern Brazil (21-34 S).

Fisheries Research 52:99-112.

Sargent, J. R., H. C. Eilertsen, S. Falk-Petersen, and J. P. Taasen. 1985. Carbon assimilation

and lipid production in phytoplankton in northern Norwegian fjords. Marine

Biology 85:109-116.

Savaria, J.-Y., G. Cantin, L. Bossé, R. Bailey, L. Provencher, and F. Proust. 2003. Compte

rendu d'un atelier scientifique sur les mammifères marins, leurs habitats et leur

ressources alimentaires, tenu à Mont-Joli (Québec) du 3 au 7 avril 2000, dans le

cadre de l'élaboration du projet de zone de protection marine de l'estuaire du Saint-

Laurent. Rap. manus. can. sci. halieut. et aquat. No. 2647.

Schmelzer, I. 2001. A dynamic seascape and a disappearing seal: a biogeographical

analysis of the population ecology of the Hawaiian monk seal. PhD. University of

Guelph.

Schreer, J. F., K. M. Kovacs, and R. J. O'Hara Hines. 2001. Comparative diving patterns of

pinnipeds and seabirds. Ecological Monographs 71:137-162.

Schweder, T., G. S. Hagen, and E. Hatlebakk. 1998. On the effect on cod and herring

fisheries of retuning the Revised Management Procedure for minke whaling in the

greater Barents Sea. Fisheries Research 37:77-95.

182

Scott, W. B., and M. G. Scott. 1988. Atlantic fishes of Canada. Can. Bull. Fish. Aquat. Sci.

No. 219. University of Toronto Press; Minister of Fisheries and Oceans; Canadian

Government Publishing Centre, Ottawa. 731 p.

Sergeant, D. E., and W. Hoek. 1988. An update of the status of white whales

Delphinapterus leucas in the Saint Lawrence Estuary, Canada. Biological

Conservation 45:287-302.

Sergio, F., I. Newton, and L. Marchesi. 2005. Top predators and biodiversity. Nature

436:192.

She, J., M. Petreas, J. Winkler, P. Visita, M. McKinney, and D. Kopec. 2002. PBDEs in the

San Francisco Bay Area: measurements in harbor seal blubber and human breast

adipose tissue. Chemosphere 46:697-707.

Shelton, P. A., W. G. Warren, and G. B. Stenson. 1997. Quantifying some of the major

sources of uncertainty associated with estimates of harp seal prey consumption. Part

II: Uncertainty in consumption estimates associated with population size, residency,

energy requirement and diet. Journal of Northwest Atlantic Fisheries Science

22:303-315.

Sheridan, M. A. 1994. Regulation of lipid metabolism in poikilothermic vertebrates.

Comparative Biochemistry and Physiology Part B: Molecular & Integrative

Physiology 107:495-508.

Silverman, E. D., R. R. Veit, and G. A. Nevitt. 2004. Nearest neighbours as foraging cues:

information transfer in a patchy environment. Marine Ecology Progress Series

277:25-35.

Sinclair, A. F., and S. A. Murawski. 1997. Why have groundfish stocks declined? Pp. 71-93

in J. Boreman, B. S. Nakashima, J. A. Wilson, and R. L. Kendall, editors. Northwest

Atlantic groundfish: perspectives on a fishery collapse. American Fisheries Society,

Betheseda, MD.

Skerratt, J. H., P. D. Nichols, T. A. McMeekin, and H. Burton. 1995. Seasonal and inter-

annual changes in planktonic biomass and community structure in eastern

Antarctica using signature lipids. Marine Chemistry 51:93-113.

Smayda, T. J., D. G. Borkman, G. Beaugrand, and A. Belgrano. 2004. Responses of marine

phytoplankton populations to fluctuations in marine climate. Pp. 49-57 in N. C.

183

Stenseth, G. Ottersen, J. W. Hurrell, and A. Belgrano, editors. Marine Ecosystems

and Climate Variation - The North Atlantic. A Comparative Perspective. Oxford

University Press, London, UK.

Smith, R. J. 1999. The Lac des Loups Marins Harbour Seal, Phoca vitulina mellonae Doutt

1942: Ecology of an isolated population. PhD. University of Guelph, Guelph, ON.

225 p.

Smith, R. J., K. A. Hobson, H. N. Koopman, and D. M. Lavigne. 1996. Distinguishing

between populations of fresh- and salt-water harbour seals (Phoca vitulina) using

stable-isotope ratios and fatty acid profiles. Canadian Journal of Fisheries and

Aquatic Sciences 53:272-279.

Smith, S. J., S. J. Iverson, and W. D. Bowen. 1997. Fatty acid signatures and classification

trees: new tools for investigating the foraging ecology of seals. Canadian Journal of

Fisheries and Aquatic Sciences 54:1377-1386.

Smith, S. J., S. J. Iverson, and W. D. Bowen. 1999. Reply: fatty acid signatures and

classification trees: new tools for investigating the foraging ecology of seals.

Canadian Journal of Fisheries and Aquatic Sciences 54:1377-1386.

Smith, T. G., D. J. St. Aubin, and J. R. Geraci. 1990. Advances in research on the beluga

whale, Delphinapterus leucas, Can. Bull. Fish. Aquat. Sci. No. 224. 206 p.

Soriguer, F., F. Moreno, G. Rojo-Martinez, E. Garcia-Fuentes, F. Tinahones, and Gomez-

Zumaquero. 2003. Monounsaturated n-9 fatty acids and adipocyte lipolysis in rats.

British Journal of Nutrition 90:1015-1022.

Speake, B. K. 2004. Comparison of the fatty-acid compositions of prey items and yolks of

Australian insectivorous scincid lizards. Journal of Comparative Physiology B

174:393-397.

Sprecher, H. 2000. Metabolism of highly unsaturated n-3 and n-6 fatty acids. Biochimica et

Biophysica Acta 1486:219-231.

Springer, A. M., J. A. Estes, G. B. van Vliet, T. M. Williams, D. F. Doak, E. M. Danner, K.

A. Forney, and B. Pfister. 2003. Sequential megafaunal collapse in the North Pacific

Ocean: An ongoing legacy of industrial whaling? Proceedings of the National

Academy of Sciences, USA 100:12223-12228.

184

Squires, H. J. 1996. Decapod crustaceans of Newfoundland, Labrador and the Canadian

eastern Arctic. Canadian Man. Rep. Fish. Aquat. Sci. No. 2359. 234 p.

Staniland, I. J. 2002. Investigating the biases in the use of hard prey remains to identify diet

composition using Anarctic fur seals (Arctocephalus gazella) in captive feeding

trials. Marine Mammal Science 18:223-243.

Staniland, I. J., and D. Pond. 2004. Variability in milk fatty acids: recreating a foraging trip

to test dietary predictions in Antarctic fur seals. Canadian Journal of Zoology

82:1099-1107.

Staniland, I. J., and D. W. Pond. 2005. Investigating the use of milk fatty acids to detect

dietary changes: a comparison with faecal analysis in Antarctic fur seals. Marine

Ecology Progress Series 294:283-294.

Stenson, G. B., M. O. Hammill, and J. W. Lawson. 1997. Predation by harp seals in

Atlantic Canada: Preliminary consumption estimates for Arctic cod, capelin and

Atlantic cod. Journal of Northwest Atlantic Fishery Science 22:137-154.

Stevick, P. T., J. Allen, P. J. Clapham, N. Friday, S. K. Katona, F. Larsen, J. Lien, D. K.

Mattila, P. J. Palsbøll, J. Sigurjonsson, T. D. Smith, N. Øien, and P. S. Hammond.

2003. North Atlantic humpback whale abundance and rate increase after four

decades after protection from whaling. Marine Ecology Progress Series 258:

263-273.

Stott, A. W., E. Davies, and R. P. Evershed. 1997. Monitoring the Routing of Dietary and

Biosynthesised Lipids Through Compound – Specific Stable Isotope (13C)

Measurements at Natural Abundance. Naturwissenschaften 84:82-86.

Svenning, M.-A., S. E. Fagermo, R. T. Barrett, R. Borgstrøm, W. Vader, T. Pedersen, and

S. Sandring. 2005. Goosander predation and its potential impact on Atlantic salmon

smolts in the River Tana estuary, northern Norway. Journal of Fish Biology 66:924-

937.

Swain, D. P., and A. F. Sinclair. 2000. Pelagic fishes and the cod recruitment dilemma in

the Northwest Atlantic. Canadian Journal of Fisheries and Aquatic Sciences

57:1321-1325.

Takaichi, S., K. Matsui, M. Nakamura, M. Muramatsu, and S. Hanada. 2003. Fatty acids of

astaxanthin esters in krill determined by mild mass spectrometry. Comparative

185

Biochemistry and Physiology Part B: Biochemistry & Molecular Biology 136:317-

322.

Thacker, W. C., and R. Lewandowicz. 1997. A comparison of low-dimensional

representations of sea-surface temperature anomalies in the North Atlantic.

International Journal of Climatology 17:953-967.

Theimann, G. W., S. M. Budge, W. D. Bowen, and S. J. Iverson. 2004. Comment on Grahl-

Nielsen et al. (2003) ‘Fatty acid composition of the adipose tissue of polar bears and

of their prey: ringed seals, bearded seals and harp seals’. Marine Ecology Progress

Series 281:297-301.

Therriault, J.-C., editor. 1991. The Gulf of St. Lawrence: small ocean or big estuary? Can.

Aquat. Sci. Fish. Spec. Pub. 359 p.

Thiemann, G. W., S. M. Budge, W. D. Bowen, and S. J. Iverson. 2004a. Comment on

Grahl-Nielsen et al. (2003) ‘Fatty acid composition of the adipose tissue of polar

bears and of their prey: ringed seals, bearded seals and harp seals’. Marine Ecology

Progress Series 281:297-301.

Thiemann, G. W., S. M. Budge, and S. J. Iverson. 2004b. Determining blubber fatty acid

composition: a comparison of in situ direct and traditional methods. Marine

Mammal Science 20:284-295.

Tieszen, L. L., T. W. Boutton, K. G. Tesdahl, and N. A. Slade. 1983. Fractionation and

turnover of stable carbon isotopes in animal tissues: Implications for δ13C analysis

of diet. Oecologia 57 :32-37.

Tinoco, J. 1982. Dietary requirements and functions of alpha-linolenic acid in animals.

Progress in Lipid Research 21:1-45.

Tjonneland, A., K. Overvad, E. Thorling, and M. Ewertz. 1993. Adipose tissue fatty acids

as biomarkers of dietary exposure in Danish men and women. American Journal of

Clinical Nutrition 57:629-633.

Tocher, D. R. 2003. Metabolism and functions of lipids and fatty acids in teleost fish.

Reviews in Fisheries Science 11:107-184.

Tollit, D. J., A. D. Black, P. M. Thompson, A. Mackay, H. M. Corpe, B. Wilson, S. M. van

Parijs, K. Grellier, and Parlane. 1998. Variations in harbour seal Phoca vitulina diet

186

and dive-depths in relation to foraging habitat. Journal of Zoology, London

244:209-222.

Trites, A. W. 2003. Food webs in the ocean: who eats whom and how much? Pp 125-141 in

Responsible Fisheries in the Marine Ecosystem. FAO.

Trites, A. W., V. Christensen, and D. Pauly. 1997. Competition between fisheries and

marine mammals for prey and primary production in the Pacific Ocean. Journal of

Northwest Atlantic Fisheries Science 22:173-187.

Trites, A. W., A. P. Coombs, and E. L. Bredesen. 2004. Whales, whaling, and ecosystem

change in the Antarctic and Eastern Bering Sea: insights from ecosystem models.

Pp: 85-139 in Investigating the roles of cetaceans in marine ecosystems, Venice, 28-

31 January 2004.

Trites, A. W., and C. P. Donnelly. 2003. The decline of Steller sea lions Eumetopias

jubatus in Alaska: a review of the nutritional stress hypothesis. Mammal Review

33:3-28.

Underwood, A. J. 2000. Importance of experimental design in detecting and measuring

stresses in marine populations. Journal of Aquatic Ecosystem Stress and Recovery

7:3-24.

Underwood, A. J., M. G. Chapman, and S. D. Connell. 2000. Observations in ecology: you

can't make progress on processes without understanding the patterns. Journal of

Experimental Marine Biology and Ecology 250: 97-115.

Veefkind, R. J. 2003. Carbon isotope ratios and composition of fatty acids: Tags and

trophic markers in pelagic organisms. PhD. University of Victoria. 272 p.

Venables, W. N., and B. D. Ripley. 1999. Modern applied statistics with S-PLUS, 3rd ed.

Springer, New York. 501 p.

Vladykov, V. D. 1935. Liste des poissons recueillis pendant l'ete 1934 par la Station

biologique du St-Laurent, dans la region de Trois-Pistoles, P.Q. Le Naturaliste

Canadien 62:77-82.

Vladykov, V. D. 1946. Études sur les mammifères aquatiques. IV.--Nourriture du

Marsouin Blanc ou Béluga (Delphinapterus leucas) du fleuve Saint-Laurent.

Contribution No. 17, Departement des Pecheries, Quebec. 132 p.

187

Vladykov, V. D. 1981. Studies of aquatic mammals. IV. Food of the white [whale] or

beluga (Delphinapterus leucas) of the St. Lawrence River. Can. Transl. Fish. Aquat.

Sci. No. 4723. 184 p.

Volkman, J. K., S. M. Barrett, S. I. Blackburn, M. P. Mansour, E. L. Sikes, and F. Gelin.

1998. Microalgal biomarkers: a review of recent research developments. Organic

Geochemistry 29:1163-1179.

Walton, M., and P. Pomeroy. 2003. Use of blubber fatty acid profiles to detect inter-annual

variations in the diet of grey seals Halichoerus grypus. Marine Ecology Progress

Series 248:257-266.

Walton, M. J., R. J. Henderson, and P. P. Pomeroy. 2000. Use of blubber fatty acid profiles

to distinguish dietary differences between grey seals Halichoerus grypus from two

UK breeding sites. Marine Ecology Progress Series 193:201-208.

Wanless, S., M. P. Harris, P. Redman, and J. R. Speakman. 2005. Low energy values of fish

as a probable cause of a major seabird breeding failure in the North Sea. Marine

Ecology Progress Series 294:1-8.

Werner, F. E., A. Aretxabaleta, and K. P. Edwards. 2004. Modelling marine ecosystems

and their environmental forcing. Pp 33-46 in N. C. Stenseth, G. Ottersen, J. W.

Hurrell, and A. Belgrano, editors. Marine Ecosystems and Climate Variation - The

North Atlantic. A Comparative Perspective. Oxford University Press, London.

Westgate, A. J., and K. A. Tolley. 1999. Geographical differences in organochlorine

contaminants in harbour porpoises Phocoena phocoena from the western North

Atlantic. Marine Ecology Progress Series 177:255-268.

Wetzel, D. L., and J. E. I. Reynolds. 2004. Definitive identification of fatty acid

constituents in marine mammal tissues. Canadian Journal of Zoology 61:554-560.

Whitehead, H., C. D. MacLeod, and P. Rodhouse. 2003. Differences in niche breadth

among some teuthivorous mesopelagic marine mammals. Marine Mammal Science

19:400-406.

Williams, T. M., L. A. Fuiman, M. Horning, and L. S. Davis. 2004. The cost of foraging by

a marine predator, the Weddell seal Leptonychotes weddellii: pricing by the stroke.

Journal of Experimental Biology 207:973-982.

188

Wilson, S. C., G. J. Pierce, C. M. Higgins, and M. J. Armstrong. 2002. Diet of the harbour

seals Phoca vitulina of Dundrum Bay, north-east Ireland. Journal of the Marine

Biological Association of the United Kingdom 82:1009-1018.

Winship, A. J., A. W. Trites, and D. A. S. Rosen. 2002. A bioenergetic model for

estimating the food requirements of Steller sea lions Eumetopias jubatus in Alaska,

USA. Marine Ecology Progress Series 229:291-312.

Wirth, T., and L. Bernatchez. 2003. Decline of North Atlantic eels: a fatal synergy?

Proceedings of the Royal Society of London B 270:681-688.

Witman, J. D., R. J. Etter, and F. Smith. 2004. The relationship between regional and local

species diversity in marine benthic communities: a global perspective. Proceedings

of the National Academy of Sciences, USA 101:15664-15669.

Woodward, G., B. Ebenman, M. Emmerson, J. M. Montoya, J. M. Olesen, A. Valido, and

P. H. Warren. 2005. Body size in ecological networks. Trends in Ecology &

Evolution 20:402-409.

Worthy, G. A. J., and A. G. Abend. 1997. Impact of killer whale predation on harbor seals

in Prince William Sound: a preliminary assessment of diet using stable isotope and

fatty acid signature analysis on blubber biopsies, Exxon Valdez Oil Spill Restoration

Project Final Report (Restoration Project 96012A-2). National Oceanic and

Atmospheric Administration, Seattle, Washington. 26 p.

Yodzis, P. 2001. Must top predators be culled for the sake of fisheries? Trends in Ecology

& Evolution 16:78-84.

Yunker, M. B., R. W. Macdonald, D. J. Veltkamp, and W. J. Cretney. 1995. Terrestrial and

marine biomarkers in a seasonally ice-covered Arctic estuary—integration of

multivariate and biomarker approaches. Marine Chemistry 49:1-50.

Zamon, J. E. 2001. Seal predation on salmon and forage fish schools as a function of tidal

currents in the San Juan Islands, Washington, USA. Fisheries Oceanography

10:353-366.

Zhukova, N. V., and N. A. Aizdaicher. 1995. Fatty acid composition of 15 species of

marine microalgae. Phytochemistry 39:351-355.

Zwanenburg, K., D. Bowen, and A. Bundy. 2002. Decadal changes in the Scotian Shelf

large marine ecosystem. Pp. 105-150 in K. Sherman and H. R. Skjoldal, editors.

189

Large marine ecosystems of the North Atlantic: changing states and sustainability.

Elsevier, Amsterdam.

190

ANNEXES

191

Annexe I. Examination of extra fish specimens for digestive tract contents. Species Region1 Month TL2 (mm) Wt.3 (g) Digestive tract contents

Am. shad UE Aug. 111 11 insect larva, insect eyes Am. shad UE Aug. 117 10.5 insect larva, insect eyes Am. shad UE Aug. 112 11.5 empty Am. shad UE Aug. 110 9 empty

Sand lance NG April 100 N/A copepods Sand lance NG April 140 N/A copepods

Barracudina LE May 242 25 1 whole krill, 6 pairs eyes (T.r.)4 Barracudina LE May 250 25.5 10 krill, 5 pairs eyes, (T.r.) Barracudina LE May 242 26.5 14 whole krill (T.r.) Barracudina LE May 256 28.5 8 krill Barracudina LE May 256 27.5 15 krill (T.r.), 1 krill (M.n.)5 Barracudina LE July 254 23.6 1 krill (M.n.) Barracudina LE July 280 29 11 krill (M.n.) Barracudina LE July 245 26.3 7 krill (M.n.) Arctic cod NG July 110 6 large copepods (Euchaeta) Arctic cod NG July 102 5.5 a few large copepods (Euchaeta) Arctic cod NG July 160 24 1 Pandalus shrimp Snailfish LE May 185 29.8 krill (full) Snailfish NG Aug. 156 66.7 amphipods (not themisto) Poacher LE July 205 N/A 1 krill, many copepods Poacher LE July 175 N/A 1 isopod, many copepods Eelpout NG Aug. 425 458 1 brittle starfish Eelpout NG Aug. 246 N/A crab legs Tomcod UE May 146 22.7 copepods Tomcod UE May 146 24 copepods Tomcod UE May 134 18.5 copepods Tomcod UE May 140 21.5 green detritus Tomcod UE May 136 21.5 green detritus, nereids Tomcod UE May 142 22.5 copepods, nereids Tomcod UE May 222 66.5 green detritus Tomcod UE May 210 74.5 nereids, copepods Tomcod UE May 187 46 1 crangon Tomcod UE May 189 42.5 1 gammarid, 1 nematode, copepods

Tomcod UE May 204 68.5 detritus, 1 small gammarid 1 UE = Upper Estuary, LE = Lower Estuary, NG = Northern Gulf 2 TL = Total length of fish specimen 3 Wt. = Wet weight of fish specimen 4 T.r. = Thysanoessa raschii (krill) 5 M.n.= Meganyctiphanes norvegica (northern krill)

192

Species Region1 Month TL2 (mm) Wt.3 (g) Digestive tract contents Tomcod UE May 215 79 1 capelan (gravid) Tomcod UE May 181 40 1 crangon Tomcod UE May 203 84 2 crangons, 1 gammarid, 1 nereid Tomcod UE May 180 46.5 2 crangons Tomcod UE May 182 42 1 crangon, 1 large amphipod Tomcod UE May 177 36.5 40 small amphipods, 1 small nereid Tomcod UE May 210 78 1 capelan, 3 mysids, 1 crangon Tomcod UE May 187 57 1 capelin Tomcod UE May 187 54.5 5 copepods, nereids Tomcod UE May 192 61.5 1 capelin Tomcod UE May 186 55.5 20 copepods, 1 amphipod Tomcod UE May 181 46 40 copepods, 1 nereid, 1 crangon Tomcod UE Aug. 182 45 1 gammarid Tomcod UE Aug. N/A N/A 2 gammarids, nereid pincers, caddisfly Tomcod UE Aug. N/A N/A 1 large gammarid Tomcod UE Aug. N/A N/A 1 large nereid Tomcod UE Aug. N/A N/A 4 gammarids, 1 small nereid, 1 slug Tomcod UE Aug. 167 25.5 1 gammarid Tomcod UE Aug. 151 25 green detritus, large nereid Tomcod UE Aug. 151 24.5 1 large, small gammarids; 1 small nereid Tomcod UE Aug. 160 30 1 nereid, few small gammarids Tomcod UE Aug. 156 23.5 1 small gammarid

Smelt UE May 156 24.5 empty Smelt UE May 140 17.8 copepods Smelt UE May 136 14.5 copepods Smelt UE May 145 8 copepods Smelt UE Aug 148 14.7 empty Smelt UE Aug 145 18.5 1 crangon, few copepods Smelt UE Aug 134 13.5 empty Smelt UE Aug 126 12 empty Smelt UE Aug 141 14.5 1 crustacean, 1 krill (M.n.) Smelt UE Aug 136 13.5 1 large, few small gammarids Smelt UE Aug 131 14.5 empty Smelt UE Aug 145 19 gammarid, nereid, themisto, insect Smelt UE Aug 125 12.5 empty Smelt UE Sept. 190 36 empty, yellow eggs Smelt UE Sept. 179 33.5 empty, yellow eggs Smelt UE Sept. 194 46 empty Smelt UE Sept. 144 19 small gammarids & copepods, eggs Smelt LE Oct. 147 23.5 empty, yellow eggs Smelt LE Oct. 175 32 many small gammarids Smelt LE Oct. 165 28.5 many small gammarids Smelt LE Oct. 166 36 small gammarids

Smelt LE Oct. 108 7 empty

193

Species Region1 Month TL2 (mm) Wt.3 (g) Digestive tract contents Smelt LE Oct. 121 11.5 few small gammarids, yellow eggs Smelt LE Oct. 123 9 1 gammarid, 1 fly Smelt LE Oct. 119 13.5 1 large, some small gammarids Smelt LE Oct. 118 9.5 4 gammarids, yellow eggs Smelt LE Oct. 105 7 empty

194

Annexe II

List of available specimens for species and spatiotemporal comparisons. Sampling of species groups by region, year, and season (1 = winter, 2 = spring, 3 = summer, 4 = autumn). L = larger specimens (> 25 cm). Subregion Upper Estuary Lower Estuary Northern Gulf

Year 2000 2001 1999 2000 2001 2002 2001 2002

Season 2 3 4 2 3 4 2 1 2 3 4 2 3 4 2 2 3 4 2 3

Arctic cod 4

argid shrimp 12

Atlantic cod 5 9

Atlantic poacher 8 4

barracudina 12 6

bobtail squid 7

capelin 11 23 7 12 30 6 6

clamworm 6 6

copepods 5 7

crangon shrimp 11 6 6 6

daubed shanny 6

eel 14

eelpout 1 7

eualid shrimp 6 3 6 3

gammarids 4

Greenland halibut 9

Greenland halibut L 4 6 1 6

haddock 2

hagfish 12

herring 12 12 12 24 12 12

herring-sardine 6 12 12 12

hyperiid 3 5 3 5

krill 5

lampreys 1 1 1

longfin hake 6

lumpfish 9 3

marlin-spike 6

monkfish 4

moustache sculpin 9

northern krill 12 6 6 3 6

northern shrimp 6 6

ocean pout 1

195

Subregion Upper Estuary Lower Estuary Northern Gulf

Year 2000 2001 1999 2000 2001 2002 2001 2002

Season 2 3 4 2 3 4 2 1 2 3 4 2 3 4 2 2 3 4 2 3

cctopus 5

plaice 5 9 10 4 2 8 2

pollock 3 1 2

smelt 12 12 9 24 36 6 16 12

redfish 12

rockling 4 3

salmon 3

salmon-smolt 12

sand lance 12 12 6 6 6

scampi shrimp 6

shad 1

shad-sardine 12

shorthorn sculpin 6 1 1

silver hake 3 3

snailfish 6 1 5 6

snow crab 6

shortfin squid 6

staghorn sculpin 1 6

striped shrimp 6 3 6 3

thorny skate 5 6 1

toad crab 1

tomcod 9 12 12

tomcod L 6

waved whelk 6 6 12

whelk (Colus sp.) 2

white hake 3 4

whitefish 2 2 1

winter flounder 4 11 12 12 1 1 2

witch flounder 5

wolffishes 9

yellowtail flounder 1

196

Annexe III

Extra figures from Chapter 2 (Prey base)

197

Figure III a). Cumulative mean abundance for 25 fatty acids as ranked by species.

198

Figure III b). Scatterplot of all scores on PC Factors 1 and 3 when weighted by total lipid

content. Labels indicate examples of three distinctive groups (scoring apart

from majority) that also had high lipid content.

199

Figure III c) Full matrix for PC 1 through 4, showing individual scores of capelin only.

Subregions were for SI = Sept-Iles, NS = North Shore and SS = South Shore

of Lower Estuary, respectively, and UE = Upper Estuary. Seasons were

spring (April-June), summer (July-August), and autumn (Sept.-Nov.).

200

Figure III d) Biplot of first two Discriminant Functions classifying specimens of common

species from the spring of 2001 into two groups: PELAGIC (smelt, capelin,

herring, and turbot) with green markers, and DEMERSAL (tomcod,

flounder, and crangon shrimp) with red markers. Circles indicate 95%

confidence intervals for group means. Shading indicates notable

misclassified specimens (tomcod = Y and flounder = open squares).

Figure III e) Biplot of the first two discriminant functions classifying autumn samples

(2000-2001) into four composite species groups when using the set of 16

fatty acid variables. The groupings were for flatfishes (flounder and plaice:

open squares), marine shrimps (eualids and pandalids: +), pelagic fishes

(smelt, herring, and capelin: small squares) and crangons (estuarine sand

shrimp: z). Shading indicates the spread of samples from the group mean

scores in the darker circles (= 95% confidence intervals). Also plotted were

202

those species sampled in lesser abundance and not used to calculate the

functions (greyed-out symbols). 12 of the 16 logratio fatty acid variables

used as input variables are labelled with their relative projected

contributions.

203

Annexe IV Alternate presentations of figures in Chapter 3

Figure IV a) Alternate view of Fig. 3.5a, with outlines for three males and one newborn

beluga (1), and harbour seal neonates (2).

Figure IV b) Alternate view of Fig. 3.6b, for a “traditional’ presentation as seen in most

publications. All individual scores are plotted on the first 2 DFs (canonical

variates), without projections of fatty acid variables.

206

Figure IV c) Second alternate view of Fig. 3.6b, with outlined boxes to suggest nested

‘metapatterns’ of groupings discriminating belugas (1) vs. seals (2), then

hooded vs. other seals (3) along the first DF, then harbour seals vs. greys

and harps along the second DF (4).

207

Figure IV d) Third alternate view of Fig. 3.6b when using three DFs, or canonical

variates. The third DF (Canon[3]) enhanced discrimination of hooded seals

and grey seals, and to a lesser extent, harp seals or harbour seals. Belugas

and hooded seals are represented here by the larger cubes relative to the other

species.